|The plan/newsfeed/web log/journal has been restored
I have made a lot of changes to the contact page to ensure that it is current in terms of links to Twitter, Facebook, Github, Google+, active development to can be tracked there. This site will be updated with the stable releases, not every intermediate development release, so I encourage you to follow the software projects which interest you on the Butbucket account, feel free to fork and send patches!
The Twitter feed contains political ramblings as well as software announcements, I may set up a separate Twitter feed for the site at some point.
|I am currently experiencing some problems trying to compile PostgreSQL for FreeBSD,
it seems anything I try to compile segfaults since I did a portupgrade -a,
as the FreeBSD project still supports FreeBSD 6, I don't think it should be the new
ports tree has broken anyway but I don't know. I tried to reboot the NFS server,
normally this is fine but this time Cannikin didn't like it and crashed because it
couldn't write to swap quickly enough; fair enough. Cannikin is now back online
briefly but I am shutting it down again to see if I can migrate Badger's hard drive
into Ivy-Mike to save some further power in the bunker. Mail, web etc will not
be accessible during these changes. I have previously announced this outage by
phone to the customer base. I will not be contactable until the maintenance window
is complete, which I am hoping will be 15:00 at the latest.
|I have just returned from the OpenSIPS (Former OpenSER) conference in Bucharest, Romanial.
It's an advanced SIP-focused UDP packet routing and re-writing engine. We have
successfully written a test module and once we have a need to write a module we should
be able to roll one out fairly quickly. Unfortunately, I lost a lot of my notes =(
Additionally, I have converted my main server - Ivy-Mike to ESX 3.5i which is now
available on VMWare's site for free. The rising cost of electricity is forcing me to
virtualise a lot of services and as soon as I can find a cheap way to get the 320 GB
main storage disk, which is ATA to work with SCSI, I shall be virtualising the main
NFS server - Badger which runs CentOS. I am hoping the energy companies are punished for
their greed. There is talk of a windfall tax. I doubt it will do any good but at least
the profits will go to the elderly who will otherwise perish this winter. Since I last
wrote to this site it would appear the political situation has got even worse =(
If you can see this, it means my conversion away from a .plan flat file and into my
MySQL database was successful thanks to the power of Perl.
|I am now looking into some more of the back-end problems and handling esculation,
rather than directly answering tickets.
|I'm currently designing some automation to pick up some issues in an automated way.
|Whilst moving equipment, Cyathus has rebooted. We suspect this is due to a possibly
loose power cable. Please accept our apologies for this interrruption. Cyathus
should be back online within the next fea minutes.
|Kernville remains down depite my badgering of the co-location facility in the USA.
Cyathus is handling mail server requests, everything else remains down. It is possible
you may see intermittent access to the mail servers because I am currently changing
the addressing scheme and installing a new router in the Bristol facility, which meant
I could have done with the co-location working!
|Kernville has gone offline. This will cause some problems with email connectivity and
really break just about everything.
21:51, I have received an update about this issue:
An update for you, At approximately 14:00 our 2 VPS nodes lost their network connection
for an unknown reason, We rebooted both of them, Our Athlon X2 regained its connection
however our Core2 node which your VPS is hosted on did not, The techs at the datacentre
are working along with an expert in OpenVZ/HyperVM to get its connection restored.
The servers booting fine and running and the VPS are running but it wont connect to the web.
Ill update you when I know more.
|Last night I was rudely interrupted by Cyathus going absolutely bezerk and screaming at
me with a piercing alarm noise. Fortunately, your captain decided not to do the simple
method of trying a reboot. I spent a long time doing things like removing one of the
power supplies I think might be faulty and putting it back in while the system was live
(it's hot-pluggable), checking S.M.A.R.T. startus on the drives etc. Eventually I opened
the cashing over the CPUs and slammed it shut and the beast was silent and happy again.
Very odd. It's a Dell PowerEdge 6450. Let me know if you have seen this before.
|Kernville has rebooted. No explanation has been provided by my co-location
provider in the USA. Unfortunately the wrong MySQL server was set to start
on reboot which did not have access to the daybologic database, this meant that
a lot of things probably haven't been working correctly for four hours.
|There have been weird routing issues in Telehouse today, with reports of call
quality problems and Gradwell's equipment going off-line briefly for a few minutes
at a time, at least in some important racks. BEThere internet was no longer able
to route to us this afternoon. AAISP (Andrews & Arnold) have had to put some
static routes in to get their customers on the internet at all whilst they
look into BGP issues. Investigation continues. Daybo Logic Kernville went
off-line. Due to Badger using this database for uptime information, this was
not officially mailed out in an automatic mailing, but it was fairly obvious
when no email was available and a lot of scripts started failing. Once I got
home and decided to switch the primary MySQL slave to Oberon (currently the master),
I found Kernville was up and running again and acting normally. It is on my to do
list to change the layout of these slaves, so there are always at least two fast
MySQL hosts available at any given time. Maybe I will have to implement some
type of round-robin or auto-fail-over decommissioning to them. No exact plans as
|Been fighting with Apache all day for my new MySQL-based dynamic web-server.
I need a totally custom Apache built with stack protection (ProPolice,
-fstack-protector) and built against mod_suid2.so. I emailed
Hideo Nakamitsu, the author (http://bluecoara.net), who was most helpful.
Anyway, the problem turned out to be nothing to do with mod_suid2.c in the end
and I was banging my head against the wall for most of the day. It would appear
to not be possible to build a sane APR, APR-util, Apache 2.2 and PHP 5 from
source on Debian. I kept either getting errors about dlname not found and assuming
a .so file (which it then went on to day No such file or directory about because it
was not built), or it would bail out with undefined symbols when I tried to load
the resulting .so into the server. I managed to get the .so by copying
/usr/bin/libtool from the Debian distribution over libtool and then re-running the
install. The solution in the end was the run make clean, run ./configure,
and _then_ copy libtool from /usr/bin/ over the one in either /usr/local/apr/build-1/
for Apache and the base directory of the source for PHP _before_ running the make.
Then the code was built with this distro-specific libtool and finally run the
make install. Now everything is working fine.
|We are aware that Kernville, our primary MySQL slave rebooted last night,
no explanation has been provided by the co-location providers. The outage
would have been no more than 15 minutes as the time/dates of outages are now
recorded by a Perl script, during which time, new logins to email will not
have been accepted. Some other services will have been affected, however,
nothing else major has been reported. We will be restructuring our slave usage
|Thanks to Peter Gradwell, Cyathus is now back online. I am presently
migrating all virtual machines back to Cyathus from Badger. They have
been very slow on Badger and they are almost unworkable.
|Grable has unexpededly rebooted. NFS mounts have continued only temporarily
interrupted. This is likely to be related to portupgrade running in the
background and replacing something critical.
|A bad 24 hours for the project. Yesterday, Badger's power supply blew up,
I could smell the nasty smell of electronics burning. It was replaced in the
afternoon, using Dumont's power supply, which had been decommissioned prior
to turning it into a virtual machine, which is still on-going. No data has been
lost. Cimarron's slave database experienced a slight problem with a duplicate key
in a non-critical table, which was manually removed amd is now running again.
Yesterday afternoon, Kernville's hosting data centre seems to be completely
unavailable. I have raised a ticket with their techs as it is still not available
now. As Kernville holds the primary copy of the database (which is also a slave),
mail will not work. I have manually pointed the public name server to use Oberon
as a master, instead of Kernville and updated the record for sql-ro to point back
to the bunker's DSL line. I expect the mail system to become fully available
within an hour.
|The database server will be offline for up to an hour from 13:45, as we make
a redundant, read-only copy, which will be up for the time being whilst we sort
out the MySQL replication.
|gradwell.daybologic.com has been down for most of the day. Some air conditioning
units failed at Sleek. This will have affected the site and the main copy of the
|Getting quite cold outside now, which is good. The primary mail server, Kestrel
is getting pretty warm though. In fact it is so overloaded I can't test the new
whitelisting script because whatever I send gets delayed. So I have temporarily
disabled ClamAV. Now only f-prot is operating. This shouldn't make much
difference, except the backlog of mail will process much faster.
|Grable has decided to reboot itself. Sorry for the inconvenience.
NFS services would not have been available for five minutes.
|Attention, there is a serious problem with MySQL replication with UUID()s.
the UUID( ) function, which generates a universally unique identifier to tag an
object or customer record, or resource, it re-executed on a MySQL slave,
rather than it's outout being inherited. This has meant that our current
database needs to be re-synchronised again! Apologies for the inconvenience.
I have decided to reduce the size of the database first, by moving out some non-
essential, large tables into a database on cust-mysql-1 before starting the next
re-synchronisation of the database. The site remains down until I have finished
working on my brand new hosting architecture and re-launch.
|Two new servers are being provisioned to handle the loss of Cyathus,
cust-mysql-1 (internally known as Dumont), will handle all current
customer databases. sql (internally known as Oberon), will handle all
system requests, (sql-ro) will handle all read-only requests on the replicated
VPS node. Badger will serve web requests. I hope to have both systems
ready before the end of the day and get some affected mail services working
again. Both MySQL servers will use an encrypted LVM to ensure customer data
is safe against theft in case of intrusion into our Bristol data centre.
|Cyathus and all VPSes have been down since last night. Engineers continue
to work on the problem to try to force the machine to boot and ensure the
problem does not continue to happen.
09:37 - A decision has ben made due to the unreliability of the drives,
and because Cyathus is too old to boot from USB, to move all operations to
Badger. This is not really desireable but must be done in order to ensure
that services become available for customers. We will be looking at
reviewing this situation within six months.
|Due to illness, I have postponed the file system upgrades on Trinity,
which I will be doing a lil bit later on.
|09:00 - My router appears to have had some difficulties, it appeared active
but had no valid routes, I have rebooted it. I have been working on a new
dlagent, it is not yet active, but will offer whitelisting functionality.
Tomorrow evening at 18:30 BST + (UTC+0100), Trinity will be offline for
approximately one hour, whilst the file systems are upgraded, because we
are about to run out of space for the main system partition.
|Whitelists are still not being used by the mail server I'm afraid,
although I am planning for this to be fixed this weekend. I am setting
up a MySQL slave on a Gradwell-hosted VPS. Although it is only a slave,
it will be the primary place for all automatic accesses to the database,
to ensure that it is highly available. Should the primary copy here
go down, most operations will still work, although it will not be possible
to change anything.
|We have now made the switch from Turf to Trinity for inbound ssh
connections. Trinity is now our main customer shell server.
It has a tool on it called dlwl you can use to manipulate your email
whitelists, although the whitelists are for the moment, ignored.
Use of the whitelists is expected in a couple of days.
|Due to a power failure earlier today, Cyathus is not available,
this is an important host which will handle all VPSes in the future,
so we urgently need to investigate why Cyathus is not happy mounting
NFS on boot and running DHCPd reliably. The results of this investigation
will be published on Friday night.
|Turf and Oberon are to be be decommissioned by the end of the month,
Oberon's replacement will be Fisher, which is a direct replacement.
Turf's replacement will be Trinity, which will become the main login
shell for external users. Due to new space being needed for the NFS
mounts which will be used for the VPSes, Cyathus' file system is
currently being migrated from home2 to home3. As soon as migration
is complete, the VPSes will be initialised.
|We have had to restart Cyathus because of some sort of obscure loopback
problem. All hosted virtual private servers will have to restart.
I could not gain root access to manually hibernate them. All systems
should be running normally within a few minutes.
|Some mail form overnight appears to have been lost whilst we were
waiting for the new MySQL database to become active (waiting for the
DNS entry to update). It appears that Postfix, if it cannot find the
virtual map, does not deliver the mail. Some of it may have been
deferred, so may be delivered later. I shall update later when I am certain.
|We are currently investigating an NFS problem which means home
directories are not available from login shells. Mail delivery is
unaffected. Cyathus is being rebuilt as a Xen VPS machine.
|Our mail server is up and fully functioning.
We are currently having an intermittent hard drive failure on
Cyathus. The system will be shut down this evening and data
will be migrated. We do not expect any major service disruption.
|16:00 - The new mail server is online and appears to be partially
functioning, incoming mail is still being rejected at present.
We have now updated the DNS for it and we think IMAP should now be
working. POP3 is not working, the server recognises valid passwords
but says the mailbox is in use. Reason unknown at this time.
21:09 - The above problem is now resolved, dovecot.index was corrupt
in my own inbox, other users were not affected. All mailboxes thought
to be active have been migrated and we are now fully testing the logic
for the incoming mail. New mail will not be processed until tomorrow.
|In order to prepare for increases in capacity, the mail is currently
being migrated to a new system - Kestrel, and is currently accessible.
Mail which has already been delivered to you will wait on our in-bound
mail servers until the mailboxes are accessible again. We hope to
restore access tomorrow by around lunch time.
|18:02 - There has been a power cut and Cyathus, Badger and Turf have
failed to recover properly. We will be investigating this fully in#
an emergency maintenance window at 18:02. Grable has on the UPS and
was not affected, Oberon recovered properly.
|21:22 - We are currently migrating the database for our managed email
clients to our co-location at Goscomb and consequently, email will not
be accessible for about an hour.
|We are currently aware of some network instability and are urgently
investigating. Neither of our internal name servers are currently
functioning correctly and consequently the MySQL server will not start.
|Attention, this is an advanced notice of scheduled maintenance and
should not be ignored. We have successfully migrated our external DNS
server to a managed server, co-located at Goscomb,
Kestrel [184.108.40.206] will now be managing DNS, which should improve
the reliability of our services slightly but mainly give us more
opportunuty to change the location of services from the outside, should
the need occur. Due to these changes, internal zones changes are
currently not taking effect. This is not a major issue.
On Saturday, the following changes will be taking place which may
result in some services experiencing some minor outages which should
not last more than a few minutes at a time:
1) This site, daybologic.com/overchat.org will be moving to Kestrel,
this will allow service status announcements to become available even
if all other systems are down.
2) The internal MySQL database will be moving to a more powerful machine
3) Important databases will be moved away from Badger and will operate
If I have time, the mail service may be moved to Kestrel.
|There appears to be an issue with BIND on FreeBSD 6.2 not respecting
the listen-on directive in named.conf, which means I am having to
change the address of the internal name server. If your code is good,
you'll be not using this explicitly or you'll be using /etc/resolv.conf,
so this won't affect you! greys.gradwell.com [192.168.0.24] is no more
and the primary slave name server for internal use is now
cyathus [192.168.0.38]. Trinity has been decommissioned.
|The issue with Fisher is now partially fixed, it will not be reloading
zones, however. So any changes to DNS will not be re-loaded on the
internet-facing DNS server. This will be resolved tomorrow when a new
host will be joining our network, which will be hosted at Goscomb.
This server will take over DNS operations.
|Fisher has gone down and has decided it does not want to access network
shares any more, this may mean access to IMAP and POP3 mail servers is
slightly intermittent, although the backup DNS servers operating at
Gradwell will mean it works most of the time.
|Due to a misconfiguration in our DNS database which up until now, the
internal DNS servers have been ignoring but now have suddenly loaded
over night, the private LAN is effectively down, all services which
need to get database access, such as mail are down. I am looking into
this as a matter of urgency, however, it will not be fixed for at least
|Shtanum has now been decommissioned but may become the Trixbox server
at somepoint, and the new release of Trixbox 2.2 or later may be
installed on Cimarron. Kohocton has been decommissioned and is of
little further use. It's internal master DNS server has been moved to
Oberon because this is also running NetBSD/i386 and so the transition
was fairly efficient. Our DNS system is now fully MySQL driven and
we will soon be publishing IP addresses for use with our DNS hosting
service. I am now working on the code to run our internal DNS server,
which is slightly more difficult than just plecluding the RFC1918-
flagged IP addresses, which are automatically selected by a new script
which runs on two machines for added redundancy.
|Installation of Cyathus has gone to plan and it is now processing IRC
statistics (where were found to have been broken since April 21st),
Cyathus will become the main beast for applications with demanding RAM
requirements. It has 4 700 MHz Intel Xeon processors (Pentium III)
and 4 gigabytes of RAM. I am currently in the process of rebooting
Dumont, which will now act as a time server, as well our main MySQL
database and I am shutting down Cimarron and Shtanum. Shtanum will
become a Trixbox 2.2 and Cimarron is being used to facililitate the
installation of Shtanum. This will mean that all VoIP phone calls
will not be possible, so I have diverted calls to ensure it is still
possible to call me.
|Kohocton has been unable to boot because it was not able to mount the NFS
file servers, this is due to Greys/Trinity being too busy to answer the
DNS lookups. Kohocton/Bilby itself is a DNS server, so I have swapped the
order of the lookups in /etc/resolv.conf on Kohocton and the system has
booted up successfully.
|Yesterday's outage: An eletrical fault blew a fuse in the main UPS,
causing Grable to go offline. Emergency shutdown commenced and all
hosts were taken offline until the fault was found. Trinity and Fisher
have been moved on to the main racks and off the UPS. The UPS is now
dedicated for the main NFS server; Grable. All file systems are being
backed up presently. Kohocton is currently not accepting ssh connections,
this will be investigated tomorrow. The primary DNS servers are still
|16:30: Shoal has stopped responding and is too unreliable.
It currently has no tasks and has been decommissioned.
As Fisher's uptime has wrapped, I have decided to move it to the rack,
which is where I want everything to be, eventually. This means there
will be a brief partial external DNS outage. I have secondary DNS
hosting though, so this will just mean some initial connections are slow.
|(Thursday): Please be aware that on Saturday morning from 09:00 until
10:30 I have scheduled a maintenance window. During this time, shell
accounts (Turf) will not be available. This is to do with upgrading
Shoal and installing FreeBSD 6.1 on it. Turf is above Shoal in the
rack and it would not be practical to perform this upgrade with Turf
still running. If you have any concerns about this, please contact
me as soon as possible. Please note that services such as email,
web-hosting and NFS file access will not be affected by this upgrade.
The upgrade of Shoal will perform more redundant mail-scanning
services and should help to ensure that any future large influx of
unexpected mail does not too seriously disrupt the network. Please
note that it will not be possible to update email whitelists from the
shell during this maintenance window.
|I am proud to announce that the Anti-Virus for email (clamscan) is
now distributed and load balanced. So email is a lot quicker than
before. Moving forward, SpamAssassin and f-prot will also be load
|I just noticed that Fisher has gone over the limit of uptime days
allowable in Linux 2.4, so it thinks it has only been up five days
|Due to some sort of issue with my router, services became unavailable
during the afternoon. They were rectified at 18:15.
|22:38: Oberon is back online. It now has 4 GB of swap and is slowly
getting munching through the backlog of mail. As for the moment,
IMAP, scanning and delivery is all handled by the on machine, you may
find difficulty connecting for a couple of hours. POP3 users should
have less problems once connected.
19:41: I am still working on Oberon, the hard drive which I was
trying to use can't be recognised, so I am having to dump/restore
via the network to create a much larger swap partition.
14:30: Oberon is no longer responding. It is seriously short on swap
space and this will need addressing before I attempting to bring the
mail system back online for 20:00.
|00:17: I have been advised by Brad to run IMAP on a separate machine,
this i probably a wise first step.
20:14: I accidentally reniced a Perl script which was in an infinte loop
to -20, which brought Oberon to it's knees straight away. I have now
commented out the CheckCountry() code, which I think was causing the
problem with the infinte loop. I'm just waiting for the server load to
get back to normal now. I think that ClamAV is going to have to be
set up as a networked-service and run on a different machine, as it is
affecting the efficiency of the IMAP service and POP3 downloads may take
a while to get going.
12:35: The mail server is back online with 216 MB RAM instead of 64 MB,
Also, I have added another 20 GB disk in case it needs to be set up with
extra swap, so we can avoid another outage.
11:50: Emergency downtime:- Oberon is shutting down because it does not have
enough RAM for virus scanning a back-log of mail. I am aiming to have
this service back online by 13:00.
|I shall be adding a filtering layer into the email server (Oberon)'s
configuration today, so there could be some minimal interruption to
service. I have also now re-pointed the main incoming ssh session port
(22/TCP) to Turf. Grable is still handling FTP though. This means the
main login shell is OpenBSD/macppc (slightly obscure, I know).
|Lost some processing power today. Castle-bravo has given up the
ghost. It's been on the blink for a while. Dodgy fan, random
reboots but now it won't boot up. I've changed the ethernet cable
and the video card but to no avail. I am going to shut it down and
re-distribute it's RAM between Oberon and Nectar which are in fairly
desparate need for more resources (more so Nectar, which is having
difficulty coping with much). I am also going to get Turf back online
today, running a Linux distribution and get it processing distributed.net
Please note that there will be a bried outage at 14:00 when I add the
additional RAM to Oberon.
|I have uploaded brand new photos of all the machines on the LAN
and have written a script which collects uptimes and load
information and writes it to the database. The web-site then
displays that information. At the moment it's updated once every
five minutes. I found out that my backup script had it's read bits
cleared (not it's execute bits, just the read bits), which had meant
there had been no backups for the previous few nights. Not entirely
sure why, I have corrected it manually and will check up on the script
permissions again in a few days.
|I have done a hard reboot of the box because I couldn't ssh in.
The IP address which was abusing the server was:
|Nectar is experiencing a DoS attack and is unavailable. I am
investigating what is going on.
|I have shut down the OverChat daemon and the perproxy in order
to move services from Trinity's private /home onto the NFS mounted
home directories. This means that the contents will be backed up
and can easily be recovered in case of hardware failure (Trinity
now has nearly 500 days of uptime but is using a very old 600 MB
|There was a five minute downtime whilst I upgraded Dovecot from
1.0rc15 to 1.0 (final release). This turned an invalid argument
into a more meaningful error, saying it could not set gid 8, I
corrected this and re-set permissions on the Dovecot deliver
binary and the errors I had been having with occasional files
not being saved disappeared. This cleared a backlog in the Postfix
queue and should prevent warning messages about non-deliver going
back to senders.
|The new mail server 'Oberon' is now online and processing mail,
I accidentally killed off postfix on 'Grable' before this process
was complete. Which meant that I was not accepting incoming mail
connections. This service was fully restored this morning and the
backlog was taken from the Gradwell mail servers. The new mail
server doesn't have some of the features the old one did yet, such
as checking whitelists. I shall be adding these as quickly as
possible to ensure we don't get snowed under by spam. The actual
back end is much better than the previous one and is more scalable,
so the usernames such as 'overlord' are no longer acceptable,
the new username would be email@example.com, for example.
Passwords have been automatically generated, encrypted and added
to the database. Phone me if you need to know the passwords.
|I had set up a cron job to automatically -HUP the named on fisher,
unfortunately, I accidentally didn't pass -HUP, I passed -TERM,
so named was not running for 12 hours today. Midnight until around
noon, this would have caused slower lookups and DNS failures because
only the Gradwell nameservers would have been answering for my
|Please be aware that at mid night tonight, when my DNS servers reload,
Daybo Logic will begin the propogation of an SPF record which specifies
that only a set of pre-defined mail exchangers are allowed to send mail
from users at daybologic.com, my other domains will mirror this record,
which is being distributed as a DNS TXT RR under the SPF (RFC 4408).
Therefore, if you have an address at any domain which I own, you will
need to let me know what your mail exchangers are. Fortunately,
Tiscali publishes an SPF record, which I can include recursively into
my specification. If you receive any bounces which mention SPF, let
me know immediately. Please remember, that not everybody checks SPF
records, so problems will not immediately become apparent and it is
not likely that if I have made any mistakes, that _all_ domains will
block your outgoing mail.
|Our main NFS server (Grable) had a main hard drive which was wearing
out. This drive had only serviced about four years and was a Hitatchi
Deskstar 123 GB drive. It has now been replaced with a brand new 320
GB drive. The total capacity on this server has been increased to
600 GB. Planned maintenance went ahead, which was annouced to our
main customers and we came back online at 13:30 today. I shall now
be migrating the internal DNS server from Grable and putting the
primary DNS server on bilby [192.168.0.18] and a secondary backup
DNS server on greys [192.168.0.24]
|Overnight, I shall be migrating the current MySQL server from Grable to
Dumont, which is a dedicated database server running OpenBSD 4.0 (i386),
450 MHz, 64 MB RAM, 8 GB drive. The memory will be upgraded as and when
it becomes necessary. The sql DNS record will be updated at 06:00
tomorrow morning. The old server will be shutdown at 08:00.
|I have am now working on a new, dedicated mail server which will
be separate from Grable. The mail server, once fully tested, will
go live in a couple of weeks. It's a P450, with 64 MB of RAM
and a 20 GB drive which will do all the SpamAssassin and virus
protection checks before passing the mail on (via SMTP) to Grable.
|The NTP server has moved to pekan.daybologic.com [192.168.0.17],
ntp.daybologic.com has been updated accordingly.
|I have renamed all boxen to be named after US nuclear bomb tests.
I have replaced core1 (now known as Grable) with an AMD 64.
I have just had to reboot the OpenBSD boxen (Shtanum) because
it kept locking up when trying to access NFS shares. This did not
solve the issue. I thought it was down to it thinking the NFS server
was on core1 so I added a CNAME to that effect and it did not solve the
issue. Ultimately, I solved the issue by making OpenBSD use TCP/IP
for NFS (the default is UDP/IP).
|Core1 is continuing to reboot itself at random intervals. I still should
have a replacement board in the pipeline, hopefully. This current
situation shouldn't affect too many services but it is undesirable.
|I found out my AMD Duron 750 MHz likes to reboot when given any
significant load. I remember it did this before but I thought it was
Windows' fault. It seems more related to whether it's running a GUI
or not. Functions perfectly well as a server. Actually, as I write this,
I do remember it rebooting for no particular reason when it was under the
bed not doing very much about once a week. Fortunately Gavin at work has
got a 1GHz machine he doesn't make any use of, so he has said he'll let me
have that. That should be fine. In which case, I should be able to get
core1 up and stable tomorrow evening. Tonight I am going to switch my
modem/router with the rouge DHCP server in to to bridge mode and see if I
can get my Linux box (currently called e11-4) to do the NAT.
On second thoughts, I won't turn it into a bridge tonight, it's getting
late and I amd just reading up on unnumbered mode and editing some DNS
settings. As the network has been up and down a bit, I'm adding a bit
of a backup into it. Most of my name servers are registered through
DomainSite at the moment but I'll probably transfer them all to Gradwell
|When I when up this morning, my file transfers were timing out and
I couldn't resolve any hostnames. When I went to move the mouse on
core1, nothing happened. Then it didn't want to boot any more. After
much investigation, I found that the motherboard was more or less
fried. Sometimes it works but not at all reliable. I've had to use
core2's old 750 MHz Duron board and chip and piece it all back together.
That's what core2 is for really. That's the Ubuntu box down for the time
being. So core1 is back but probably a bit slow. It will do, means
all mail, sites etc. should be working again. I'll need to get a new
mobo for core1 as soon as possible now. Services restored 23:30
|Talk about elite, now I'm not using POP3 any more. I've merged my work
and private mail addresses with the help of fetchmail (cheers Eric),
and I've set myself up a private SSL-enabled IMAP server so that I can
view the same email from my laptop as well as any of my desktops.
Uploading the mail back from Thunderbird is taking a long time though.
I got a DVD burner working but not on FreeBSD, unfortunately. I had to
resort to using Linux. I'm setting up Ubuntu server for Jeramy in the
rack. I'll be enabling it with the use of mod_proxy to set up a reverse
proxy. I shall do the same for the Trixbox too, so that can be accessed
from the outside world.
|Advance notice of downtime from 23:00 Friday evening, which may lead
into the early hours of the morning. These changes will change my local
modem into a modem/bridge and a Linux-based router will take on the role
of of my border router. All services will be down during this time and I
hope to keep this down time to a minimum. This should result in a more
flexible service and the removal of a rouge DHCP server which cannot be
|I've installed ISC-DHCPD on a P50. An old slow computer. It works,
no problem there. However, my Actiontec DSL modem makes a DHCP
offer even though support for that has been explictly switched off!
This is a major bug in the firmware and confuses everything because
it knows about different DNS servers than the internal clients.
Absolutely ridulous, it's an old modem/router, I guess. It would appear
I'll have to think about getting something a little more advanced. I
wouldn't mind using it purely as a modem and assigning the public,
static IP address assigned to be by BulldogDSL to the external
interface of a FreeBSD box. I've decided to install Mac OS X back on
the G4, I can always run updates etc with VPN and use it as a headless
machine on the DreamStack.
|Advance notice that I expect there to be 15 minutes of downtime
between 21:00 and 21:15 on Friday 12th January (tomorrow).
I need to test out another DVD writer. Also, I have just changed
a few more names of defines in the OverChat project. Mainly to
separate further the concept of the old monolithic tree from the
separate packages which are going to be released. I have now
ditched the old OVERCHAT_VER* tags because every project will have
it's own version number. There is no one specific release version
|I shall be shutting down core1 for approximately 10 minutes some
time between 15:30 and 16:30 in order to change a DVD writer over.
The Trixbox has been offline for most of the day because it doesn't
like mounting more than one NFS share, I don't know why. Complains
of an incorrect superblock of all things! I have just upgraded it
to Trixbox 2.0.0 but I don't think this is related because I added
the second NFS mount without rebooting it. The next idea is to
add a PHP script to the main site to send all requests to it so
that I can access it from outside of the LAN without needing to
do a double-ssh login. I am also installing Gentoo on core2.
|Current plan is to write a patch for sipgrep so that I don't have to bother
entering my email address every time I want a copy of the SIP trace.
|I've now got myself a dual-layer DVD-writer. The folk at work think
it won't work on FreeBSD. The previous one doesn't. I get an I/O
error, which is why I was looking for a new one. I need it to write
off remote backup DVDs really, in case of fire. I do have a backup
drive but I would also like a dedicated remote backup DVD spindle.
I had one but I've just collected it for refreshing because the data
was two years old and a two year old backup is a pretty poor show
|I have removed the network cable from gnu.daybologic.com (The Compaq),
it is not presently in use, so I have shut the machine down temporarily.
It's the only ethernet cable I have free and I need it for use with my
new lappy. I have switched gnu off.
|I have purchased a new, used laptop. 700 MHz, 128 MB RAM and a 10 GB
internal hard drive. I had to go to Swindon to get it. Local TradeIt,
very useful. I'll be installing FreeBSD 6.1 on it with all the useful
tools for network sniffing etc. I have found under FreeBSD 6.1, for a
fresh install, there is a new trap ready for people installing it, you
now have to explicitly say which kernel you want to install and if
you're not aware of this you may find yourself with a new and useless
install with no kernel installed whatsoever. Recovering from this is
possible but difficult and you'll need the fixit floppies.
|Rebooting core1, services will be affected. My DVD burner doesn't work
and I need to do some backups to ensure continued reliability. I'm
upgrading from FreeBSD 6.0-RELEASE to 6.1-RELEASE. I keep getting
Input/output errors reports by burncd for everything I try to do, whether
it be DVDs or CDs.
|We have suffered the worst period of downtime ever. I've been able to use
the internet but that's about it. Most services have been down. Major's
root file system became corrupt and I had just taken down some of my sites
to redesign them. During the commissioning of a new server to replace
Major, (core1), I managed to install FreeBSD 6.0 on the wrong hard disk
and destroy my /servers partition. Fortunately, I had a backup but it was
about six months old. This is now what I have had to revert to. I've
renumbered and redesigned the network somewhat, cut out one layer of
routing and I have an FTP server on core1 which can cope with active and
passive clients even though it is behind NAT. I did this by restricting
the ports it is allowed to accept from clients requesting passive
transfers and port forwarding these ports to the server. I hope to have
my web-sites back online shortly and I am now running my own mail server
(Postfix). I am still running my own DNS but I am using two separate
machines for internal/external DNS. The addresses of the computers have
changed to: border-router [192.168.0.1], core1 [192.168.0.2],
e11-4 [192.168.0.3], oc0 [192.168.0.4] and transerv [192.168.0.5].
|Major to be partially decomissioned. I have decided that Major, which is
a G4 PowerMac (PPC), 400 MHz will be partially decommissioned. It is
presently Daybo Logic's primary server for nearly all services but as time
goes on, it is not really good enough for this, so it will become an Aqua
based desktop for my private use only in the future. Stage one of the
transition will be moving services onto a Debian (GNU/Linux) based box.
These will be minor services, mail, time server, this has already been
completed. Stage two will involve the custom building of a fast PC,
1.4 GHz with a lot of space and running FreeBSD. Then Major will move
all of it's data onto this PC. This PC will share the load with the
Debian box and also act as a desktop computer. Initially, before the
switch-over from Major, this PC will be used at work. Ultimately, it
will remain here but be used remotely from work.
|Started working for Gradwell.com Ltd. Used to be self-employed fixing computers.
The majority of my life has been dedicated to computers and the digital universe.
|Smith has been decomissioned. I have decided to set it up at work as my
FreeBSD 6 desktop. Thompson has been turned off. It's an OpenBSD box.
Nice... but I don't actually have a use for it. It's still connected and
ready to roll should anybody have a use for it (ie. software testing and
porting) but until such time as an actual use arises, it is turned off to
save power. Contact me if you need access again.
|I've set up my own mail server. It's fully functional. This will mean
we'll have email at overchat.org and daybologic.com (possibly other domains
too). I'll be able to host folks' email services as well as my own box.
The folks at Gradwell will have sorted me out a mail box by now. When I
get the details, I'll add them to my .fetchmailrc file and just log into my
own POP3 server here at the bunker. That will merge all my mail in one
place. I'll probably have to copy the rules and directories to my work
computer, otherwise I'll have a totally insane inbox which won't be workable.
Oh, it turns out they support plans at work on the Newred server.
I'll set up a script to update my plan from here >=)
|The address of proxy.daybologic.com has changed From 192.168.1.4 to
192.168.1.7 (Major to Compaq). Please use the cname. This is to free
up more resources on Major, which is being over-used.
|301 days uptime today. Highest ever personal achievement. Great news,
I found a new job as internet technical support.
|Attention, we have reached 292 days of uptime and I feel it is time to shut
down. I have plans which are near completion to re-wire all the computer
mains power via a bank of control switches and 5A fuse wire.
|I shall shut down Major today. Applications are randomly crashing and it
needs some routine maintenance, such as the new security updates and
Quicktime upgrade from Apple. The downtime will be from 17:30 until 18:15.
|Mad storm last night. I nearly shut down my equipment and just had it
running from the UPS. It was the middle of the night though and I was only
half awake. Fortunately, no damage has been noted to date. At work,
as the storm continued during the day (although not so bad here), they had
a serious equipment failure due to a lightning strike. Forget not the
voltage of this stuff, it can seriously cost you money. I lost a couple
of sticks of RAM (old SIMMS) in the retired machine Howard, about 9 months
ago. It's bad news.
|After my employer shut down their control centre in Bristol, I am now faced
with the prospect of finding a new job quickly. If you want somebody with
lots of technial skills and knowledge, I am looking for office work,
computer work, administration. Something along those lines. If you're
interested, please let me know, as soon as possible.
|Blake now supports MySQL, trying to get the server built to use it now,
using autoconf. Not much getting done though, unfortunately. I tend to go
to sleep as soon as I get in from work to avoid this poxy heat. Still on
track to have the car on the road next month.
|Compaq (the MySQL server) didn't want to start after I brought the systems
back up. Sorry about that. Extended down time is not a good thing.
It's not that reliable, that system, because it doesn't like coming back
from a power outage. It needs to be on a UPS really but I would need
another UPS, they are expensive. I'll have to start saving up =(
Alternatively, I could move the database server to another box.. but which
box it could run on, except Major, at the moment I am not sure. I have
fitted another switch disconnector to the side of the desk to isolate the
hi-fi and computers on the other side of the bunker circuit. The UPS
continued to power the bunker's critical servers during this operation.
This meant that 70 volts were always available on certain parts of the
circuit, even with the mains power off... it leaks into the mains via the
plug! Not a major risk, so I worked on it semi-live. My Plasta-plug
testers blew up! Quality gear... not. Anyway, don't expect much
progress... on Wednesday, we are expecting 36 degrees celcius.
I do not work in that heat! I mope! I shall publish video and photographs
of the re-wiring process soon. In OverChat's news, I am proud to annouce
that the current libblake source now reads users from the MySQL database!
|Be aware that at some point within the next 24 hours, the entire network
will shut down. Heath, Hauge and Major will still be alive because they
are served by a UPS and I will have a close eye on the clock during this
operation. I have plans to check mains wiring connections and put a new
Earth bar into both isolators. The Daybo Logic database server will be
down, so most sites won't function normally. New photos will go up when
I have finished re-wiring.
|The hot weather continues and I can't concentrate on this OverChat SQL code.
So I have shut down Thompson and Smith to try to regulate the heat.
|Yesterday, I noticed a few dried leaves blowing across the ground. One flew
inside the van, actually. I declared that Winter had begun. Let us hope
that I am correct. I brought back that AMD Ahtlon 1.4 GHz that I bought
from Iceslayer Columbia for £40. I dismantled Smith's boards and replaced
it with this new board. It is tiled slightly, it's too large for the top
shelf, so I'll have to find some wood from somewhere to hold it level.
Unfortunately, all graphics cards I have are incompatible with SecondLife.
How pedantic are they? It works on the Mac (Major) but that computer is
just too slow for the world. Sigh. So I need to get an ATI Radeon or
something but I need to get my car back on the road and I have more
important priorities at the moment. So it looks like I'll be off
SecondLife for at least three months. I ran the car engine for about
30 minutes and ran it around in the garage. I'm going to wash it this
morning. I'll get a couple more quotes before I start my insurance off
on 1st August. I've split the Blake configuration to support the following
database modules: legacy, MySQL, PostgreSQL.
|I am pleased to annouce, according to the BBC's notoriously unreliable
weather web-site, that this appears to be the final day of the heat wave.
Over the past few days, I have come home from work and slept for an hour
or so, then done some reading. I have added a user count to the OverChat
web-site so one can see how many users are registered. This work will
continue in more detail after the heat wave. I have also been thinking of
re-designing the Daybo Logic web-site to make it a bit more modern.
|I got the battery out of my beast today. Car battery, this is. I have a
CitroŽn AX (a pretty basic car). Drives like a milkfloat but it's cheap ish
to run. The battery is flat so I've taken it out to re-charge it.
It doesn't get the chance to charge up because it's off-road. It's only
done a mile since it's been offroad and that's just traveling around in the
garage. Anyroad, detaching the terminals was easy enough but there was also
a plate on the right-hand side of the battery which needed a nut loosening
to take it off. Unfortunately, there is really not enough to room to move
the spanner around in. I ended up having to use the spanner vertically and
I kept having to adjust it all the time. A stupid design . Finally, I got
the battery out of the car, now I'm going to charge it up over-night with a
battery charger I borrowed from somebody at work.
|29 deg C today, 31 tomorrow and 32 on Wednesday. Yes, the filthy weather
has engulfed us again. There is hardly any point living during this weather
and I don't expect there will be much progress on any of my projects.
Why can't we ban summer? It's not as if there is much point in it.
The economy suffers over a certain point if people are too hot to work.
There are optimum temperatures, you know? One can't type if it's extremely
cold. One can't think or be bothered so to do if it's too hot. I am not
happy about the situation and what really irritates me is these
sun-worshippers who hope we'll live in misery and foul weather all year!
My grandmother is just as bad on this line of reasoning. There is no point
being alive in this filthy weather.
|Ah, had to fix somebody's laptop. Had to buy a new multimeter. It turns
out their transformer was buggered. No power coming out of it at all.
Anyroad. I tried this multimeter on the mains supply to test amperes...
I couldn't see how it would work on anything but continuity and voltage...
it didn't =X I blew the 20 amp main fuse. All machines on the UPS
survived the outage. Main fuse replaced. Down time possibly ten minutes.
Black soot over test lead =D Don't worry lads, we're back in play.
heath.daybologic.com: 8:14PM up 188 days, 20:59, 1 user,
load averages: 1.00, 1.00, 1.00
|Phone service to the NOC has been discontinued. Unfortunately, too many
people abuse this service by constantly phoning and not leaving messages.
Those who do leave messages are not generally enquiring about anything
useful or leave the old "please phone me" rubbish, so I cannot determine
if it is worth calling them. Therefore, I have disconnected the line.
|The third-party DNS for Daybo Logic was down last night for several hours.
I was unable to receive email and a lot of my sites rely on scripts based
on this site, so those sites failed as well. All seems to be normal again
|I have written and successfully executed a Perl script to convert the
legacy OverChat binary database into a MySQL table. This allows for
significant expansion in web-based services, database size and makes it more
easily maintainable. Watch this space for online registration, statistics,
search and password recovery.
|There is a storm going on outside, so let's hope there are no power
interruptions. If the SQL server goes down, it won't come back up by
itself and the web-sites will be down all day. I need another UPS for
the other end of the NOC.. but I can't afford one just now. I will look
into it though. I should be able to get one in about three months.
|Unfortunately, for the third time within a month, Finder has failed.
I get disconnected from my NFS share during a file copy and that's it.
There is no way to recover the GUI in any meaningful way without a reboot.
It's about time Apple fixed this problem. Fortunately, they have just
released some security updates, so at least the reboot will be worthwhile.
|Finder has gone wrong again. The problem is copying large amounts of data
over NFS. If that connection goes wrong, the copy dialogue box in Finder
never recovers and never disappears. This seems to have other effects on
the system, like my inability to shut down Preview. At least I only have
one day of uptime but I must wait now until the server has finished
processing the thumbnails for my web-site so that I can safely reboot.
|Major (Mac OS X) Finder has gone wrong at last. It went wrong whilst
copying large files across NFS. I've decided to take the opportunity to
reboot it and install the security updates from Apple. Final uptime is:
9:12 up 99 days, 10:16, 5 users, load averages: 3.98 3.92 3.78
|Somebody sent me a patch against daybodep-0.5.0 to correct a crash.
Thanks very much!
|I released libblake-0.2.2 today. It has a GUI configure program for Win32.
There is a bit of a thunderstorm going on outside. I hope it doesn't
interrupt power to the bunker. Heath's main console is now taking the
keyboard's input as garbage, after being up for 138 days (manual KVM switch).
I still have SSH control of it but it seems like the beginning of the end
unless I can find a solution. Hauge is on 128 days uptime if it fails.
|Due to continued expansion of my current project, I will need to switch
Smith and Pavilion's operating systems over. This should take no more than
a week. Pavilion will be the new FreeBSD 6.0 box and Smith will be the
Windows 2000 box, with my new IEEE1394 card installed.
|At the end of the month, libblake-0.2.2 will be released, as a separate
package. I am currently working on a different project with somebody from
work. Also, from May 1st, all computers which are not doing anything will
be shut down to keep the heat in the lab down. Obviously if you need
access to them, send me an email and I will start them up for you.
This notification will end on September 31st, when all machines will be
running constantly again.
|I have released DayboCrypt 1.2.0. It now has support for SHA-256
(using code I found by Colin Percival) and has a totally re-written build
system, using GNU autoconf and automake. OverChat is using this version
of the library now, the previous version(s) called dybocryp are officially
|The OverChat site is back online. The forum is unfortunately down for the
time being. I will be introducing a new (ten year old) OpenBSD 3.8 box to
the LAN soon. OverChat was down temporarily because of another computer on
the LAN using the OverChat login server's IP address for a few hours.
I have got Reiser4 compiled into Linux 2.6.12 and I have set up TrueCrypt.
Working on new packages of DayboCrypt and libblake (which will now be
separately packaged from the rest of the OverChat tree).
|Cameron's file systems are becoming corrupt. This is a side effect of one
of the hard drives wearing out. I was trying to upgrade it to a new kernel,
with the Reiser4 patch but unfortunately, it's on it's way out. So I'll
remove the defunct hard drive and re-install the OS. I'll be backing up the
MySQL databases first, obviously. This could mean the sites are up and down
a bit this evening.
|Updated resume uploaded; OverChat 0.2.0 server (Restronguet) uploaded.
If you need an administrator (secretarial or systems) or if you have any
job offers for me, please email me!
|Power cut. All machines except the three on the UPS reboot. Power cut
lasted around 1.5 seconds (estimate). I have difficulty getting Compaq
back up. I am due to release dlscripts-0.3.0, rttb-0.0.3 and get config
files working properly on the server side of OverChat today.
|I have released dpcrtlmm-1.2.0. This time, the configure scripts have been
completely replaced with GNU autoconf scripts and the code has been
re-tested on MS-DOS.
If you are reading this and you are an Open Source programmer with plenty
of time on your hands and you're looking for something to get your teeth
stuck into... why not join me on the OverChat project. More information
can be found on http://www.overchat.org. I could really use some more
programming hands on the project. It progresses but slowly, as we need more
programmers, just like you! You need to have some experience of UNIX and
plenty of experience with C.
|The path to Restronguet continues. I have created an installation script
for dlscripts and re-released it. I went to a see this weird "music" group
with some sons of mine who were up from Kernow and it was absolutely
terrible. I must be getting old because it just wasn't music.
Cameron has started making a grinding noise! It has two hard drives,
let's just hope it is a fan bearing or something. I've made a new
spreadsheet using Gnumeric for my budget.
Update: I am going to have to reboot Major. Thunderbird and Finder are now
unusable because of defunct processes, I can't browse for any files.
I have scheduled a reboot for 21:00 tonight. Major will be down for
approximately an hour and during this time, essential maintenance will be
performed, as well a memory upgrade. Also, it is one of Cameron's hard
drives which is causing all the noise, this is very bad news. I shall make
a backup of the backup directory immediately.
|We now have three OverChat servers running. Stable, Testing and Unstable.
The latest build supports offline errors, returning code bec_user_offline
to clients who try to talk to users who are no longer online. There was a
bit of a glitch compiling because we don't know whether to link with
-lresolv or not in ocquery. I'll need to write a configure test for this.
There was also an assertion failure in OberChatten which is difficult to
re-produce but can happen when disconnecting and reconnecting to several
different servers. In three more features' time, we will release the
long-awaited OverChat 0.1.8 (Restronguet). Ask not what OverChat can do
for you; ask what you can do for OverChat! I am also planning to re-release
all Daybo Logic libraries with GNU Autoconf support to lower the barrier
to entry for support libraries.
|Finder has stopped working on Mac OS X-based Major. Logging out and
logging back in didn't work. Sending the defunct process SIGKILL didn't
work. I will just have to live without Finder for now, as I am determined
not to lose the uptime, which is currently 40 days. Maybe I'll find a
replacement file browser somewhere.
|So I bought this Commodore 16 from somebody at work... he gave me a load
of games with it but most are for other machines. I've set it up via
the RF output to a TV card in Pavilion. While fitting the TV card in
Pavilion, I managed to cut the power to Compaq somehow. Loss, 22 days
uptime. Nevermind, the highest uptimes should be on the UPS anyway.
I've been debugging OverChat but I have found some mutexes which rely on
recursive behaviour. I'll have to just support this for now and
re-design server_netbuffer later on.
|Smith rebooted. Reason unknown, it must have been just before I walked
into the NOC. No other machines rebooted, so it seems as if it is not
related to power. I also plan to reboot Pavilion tomorrow, as I have some
documents to scan. I am expecting a new release of OverChat very soon now.
I am trying to debug the current source, as one can connect but no
challenge is sent to clients.
|Daybo Logic ceased all business operations today. I am no longer
self-employed and I look forward to a new beginning for Open Source
software development and searching for an IT related job, where I can
share my knowledge with others for the benefit of the digital age.
OverChat development will continue as time permits. I no longer go out,
except when I'm working and all my spare time is spent programming or
|Twice, for no reason, the power went out this morning. I don't know
whether it was caused externally or internally because the lights
weren't on. It was about 06:30 and the UPS started beeping. Then, ten
minutes later, it started beeping again and this time Smith rebooted.
All other hosts survived.
|I found the cause of yesterday's power outage. Unfortunately, when I
opened the door to the bunker, there was an electrical flash and the
power went out. Loose wires in the main switch. Uptime on all hosts
except the three UPS backed hosts was lost. The SQL server was down for
about 15 minutes. The website would have been down during this time.
All services have been restored.
|Unfortunately, it seems we had a very short power outage today. Newly
commissioned Compaq rebooted, as well as Cameron and Smith. All UPS
backed hosts plus Pavilion, which is not battery backed, survived.
Strangely enough, Compaq booted back up without asking me to press F1.
Which is lucky, seeing as I don't have a keyboard attached.
|After the systems came back up, Hauge, now running Debian GNU/Linux Sarge
3.1 had better support for APM and allowed the BIOS to spin down the hard
drive every five minutes. A couple of minutes later it was spinning back
up again, so I had to reboot it to sort out the settings in the BIOS,
as this kind of spinning up and down all time time is not a good idea in
the long term. I see no further need for outages in the project and I am
now trying to build some serious uptime.
|The systems will be offline until I post another message. I have attempted
to set up a web-based uptime system but UDP is not supported properly by
portfwd on FreeBSD because port forwarding is not transparent. This is
supported properly under Linux. My copy of FreeBSD on the gateway is
obsolete and the system is too old for the latest versions, it seems.
Therefore, I will be switching the gateway over to Debian GNU/Linux.
Therefore, we are offline. All sites except this one will be down and
this site will only partially work. Adios!
|Attention programmers! OverChat has officially been suspended. I am
still willing to accept patches and release the current source tree on
demand. I have a much higher priority agenda going on, of finding a new
job. Don't worry, OverChat users... you will still be able to log on to
the server, there should be no problem with this. Let me know immediately
if you have a problem... otherwise, I'm using electricity and providing
a battery backup for nothing! If you're looking for a programmer or
systems administator, let me know!
|The systems will be unstable for the coming 24 hours, a couple of
short-planned shutdown events are necessary to install UPS equipment.
It is also necessary to turn off the mains power to remove an extension
cable which was directly wired to the mains. That extension cable will
now be wired to the UPS. There will be two main extension leads on the
main development workstation, one surge protected and UPS backed and one
surge protected but not backed up. I am writing an application for
my employer at the moment. It is a tariff calculator. You wouldn't
believe how many sites and deals are going on. Different tariffs and
discounts. It's actually quite a difficult problem and I've decided to
represent all their workings in a C++ class hierarchy. Still looking for
a new job, if you want to hire a systems administrator or programmer!
|Howard has been retired. The EIDE bus must have failed because a while
ago it couldn't boot from floppy disk, even after I changed the cable
and the floppy drive and then suddenly, it wouldn't boot. When I unplugged
the hard drive, it booted but when it got to the BIOS' AT attachment
detection stage, it hung. Norman at work gave me an old computer of his.
It's a P120, with 16 MB of RAM and a 1.2 GB hard drive. I have fitted
an extra 6.7 GB drive and added another 8 MB of RAM. I've installed the
latest Debian GNU/Linux testing distribution and set up mysql-server-5.0
on it. We're back in play, ladies and gentlemen.
|Howard has failed. I can't get any output from the VDU. I knew that
previously I had to ditch some memory because it was crashing and that
it can no longer boot from floppy disk, so I guess it was just a matter
of time. Howard is a router and therefore this will cause some
connectivity problems with certain computers on the LAN. I'm working
09:00 to 17:00 doing something completely different today, so I'll have
plenty of time to think about it and during that time I _may_ decide to
give up on the idea of using Howard as a router. Technically, it is
unnecessary. The only vital service Howard runs is the SQL server.
I will probably set up a SQL server on the opposite side of the room,
over a 10 Mbps link. An old 486 or something along those lines.
|I went to Witchurch to the computer fare there. I picked up a four port
passive KVM switch. So now I'll be able to have full access to all my
boxes on the main workstation >=) Also, I am available for work.
If you are looking for a programmer or systems administrator, look no
further. Email me at once!
|Replacing some taped together wiring with terminal blocks, re-doing the
Earth connections for all equipment. Will be down for an hour and a
half between 17:45 and 19:15. Most of this website will not be
|Sorry about my recent trip into obsession with electrical things. I am now
concentrating all my spare time back into the OverChat project. I have
released libdlini 0.2.0 and I am now working on stripping the buggy
threads code out of Blake and the server and combining them into a new
library which will be called daybothreads. This will improve the stability
of the server and the former release, libdlini will allow easier use of
the .ini configuration file supplied with OverChat. I expect a
stable 0.1.8 server to be released at the end of the year, after which we
will begin work on a full SQL-enabled release.
|We have done an electrical audit of all devices in the bunker. Two 13 amp
fuses were found in devices which required a 3 amp fuse. Most devices
had 5 amp fuses, when only 3 amp fuses were required. Our mains supply is
now rated at 20 amps. I am going to also pick up some thicker cable today.
This supply is drawn from two separate mains sources.
|Smith's hard drive failed. I think it may still be useful as a portable
hard drive, for carrying around a copy of the FTP server on. I have just
been and purchased a new hard drive, 80 GB. Too big for a desktop with
server but that's all they had. Cost, £44. Smith does not run any
services; OverChat was not affected. Now I must release a new version
of dlini, get a couple of boxes converted to FreeBSD 5.4 and get Smith
back up and running within a day and a half.
|Attention programmers, I am going to crank the handle at 10:00 GMT and
the power will be out to all Daybo Logic systems for up to three hours.
The usual thing, re-wiring and moving systems. This site will be of
limited use during this time.
|Announcing OverChat 0.1.8-pre1. After a long wait, Daybo Logic re-iterates
it's commitment to the OverChat project with this pre-release. After a lot
of difficult maintenance work, this release is out, however, the server
has some bugs and isn't suitable to replace the one on the stable server
just yet. I hope to turn that around as soon as possible.
|The network will be down for a while today because I am re-wiring the
Daybo Logic LAN. Everything will go through the switch-disconnector I
mentioned before... then it will go to a surge protector, then into an
extension lead to my secondary workstation area. The secondary workstation
will therefore be isolated by the first and will be based in the corner,
by my cupboard.
|The network will be down this evening, as I am fitting a new isolator for
the entire Daybo Logic LAN. It's a MEM Exel 63 amp 660 volt
|I have decided to spend a minimum of two hours a day working on the
OverChat project. A consequence of this will be the release of several
previously unreleased or under-maintained libraries. Don't expect them to
be 100% bug free, as their release is purely to support OverChat. I am
more than happy to take bug reports on them, however.
|Unfortunately, my laptop is finished. I have determined that it is either
the main board or the processor... seeing as it's obsolete anyway, there
is no point trying to revive it. I have decided not to run an FTP server
any more. All the files have been moved onto the web server.
Considering HTTP supports virtual name-based hosts it's a better solution
all round. Meanwhile, my son Dagsylad is looking to sell his old NT
boxen... and I'm in the market to buy it.
|I have set up the Squid cache on Major, with it's new MacOS X installation.
It's running as effective UID/GID nobody:nobody. I'll make a separate
proxy:proxy set up for it later. I have managed to get it to boot up
using /Library/StartupItems/. Unfortunately, I just tried to boot up my
laptop, which has been idle for a couple of months and I get no response.
The CD-ROM light stays lit and it does not attempt to access the floppy
disk drive. I have probably disabled floppy disk booting in the CMOS setup
utility. I'll have to take it into work (where I store all my tools) and
open it up. I'm going to get down there right away and see if I can debug
this. The plan was to erase NT 5 and install NT 4 to speed the machine
up without spending any money. It seems I'll probably have to spend some
money on it after all, perhaps the hard drive is damaged.
|Well, this is a post from MacOS X. I have installed Xcode and I am up
and running. Well, almost. I have NFS servers and clients re-sorted,
between all the machines. There is much still to do. Including restoring
|Attention programmers. I have purchased a two PS/2 port to one USB plug
type A convertor from Maplin this afternoon and have decided it's now or
never for the installation of MacOS X, which I have had lying around for a
while. I can finally get the G4 onto the KVM switch, so I can use it as a
desktop and a server. I have also recently purchased a
USB 2.0 PCI 2.1 card. I shall try to get this to work in the G4
(which presently has Firewire and USB 1.1 sockets). Failing that, I'll
back up over the network. Anyroad, the idea is to back the whole thing
up to the external 200 GB drive, re-format both hard drives and replace
all config files with my config files (not without due care and attention
to detail of course), once Mac OS X is installed. Fellows, OverChat and
all sites but this one are _down_ until further notice. Exciting times
|Filthy weather has engulfed England. Wake me up when it's over. Life is
not worth living during such extreme conditions. Don't expect any code
or work to be completed at all. I'm still going to work during the day
but I am just going through the motions and I don't think properly. All
I do is moan about the weather. Anything over 72 degrees ought to be
illegal. I finished my new LAN, by the way. Tried to get GNU/Hurd up
and running but I did not succeed. Perhaps I'll have another go at
cross-compiling it on FreeBSD later.
|Expect the LAN to go down at some point within the next week. Fortunately,
at least temporarily, the filthy weather has ceased. My plan, when I
have enough screws, is to dismantle the LAN, discard the cases and put all
the motherboards inside various drawers inside furniture. Bit like those
cruncher-obsessed folk do... except that my machines are old.
|Howard is unstable. I believe it to be the heat. We are having some
really filthy weather in Bristol at the moment. Very humid, 90 degrees
today. Not pleasant. Fortunately, I have the day off. Howard is a hack
up really, with the power supply loose in the case, next to the heatsink
over the CPU. Howard is continually rebooting or locking up.
My workstation and the OverChat server are behind Howard because Howard
is a router. Therefore expect connectivity problems and limited ability
to chat. the SQL server and website are unaffected. This is likely to
continue until this filthy weather ceases to trouble us.
|This has been a slightly annoying week for the LAN. First of all, the
DNS/firewall box (which has two interfaces, internal and external) stopped
routing traffic... eventually I rebooted it and found that the external
interface wasn't being detected. I didn't have time to fix it as it meant
dismantling the whole LAN to get to the box, which is under all the other
computers. I replaced the card with another card, there was nothing wrong
with the card but it's working again now. Then I found that on the other
FreeBSD box, the same thing had happened, so I replaced one of the two
3Com 100Mbps cards with a D-Link 10Mbps card and had to reboot on a DOS
disk to configure the card to use the correct interrupt and base address
settings. Anyway, that's meant that anything behind the second router
is now limited to 10Mbps access to the main server. Not good. I have
left all computer cases off for easy access. Also, I installed bad memory
into my main workstation, which corrupted the file system so badly, I have
to re-install Linux from scratch. Next week, I'll be manning a car park
in the mornings and I'll be on in front of my workstation in the
|Daybo Logic is stale again. I'm too busy working as a car park
controller. The job involves monitoring multiple car parks with CCTV
monitors. Fixing customer problems and trying to monitor crime. However,
the hours aren't that clever. It's 12 hour shifts. Days are OK but
nights are terrible and nothing happens. Fortunately, one of the
supervisors is leaving and I might be getting his job. This will involve
going around fixing machines, barriers and doing things like printing
off tickets and driving around in the van. On other issues, OverChat is up
and running but the stable and running version of the server is still
version 0.1.5 with patches. I've just released a new version of
OberChatten (0.4.6) but we still haven't got the SQL working on the server.
For the last month, I've been doing nothing but working. I should get
my sound system set up this weekend. I bought a bike from Halfords the
other day, so I won't have to keep spending on the car.
That's all for now. Not much of a plan entry really. I'll report back soon.
|Palmer has received a driving license. This changes everything. I can
now expand my operation, increase my prices to £15 / hour, carry more
stuff around and do 2-3 jobs a day instead of one if necessary. If I
choose to move, it is another good thing to add to my resume, that I can
drive! I shall spend a couple of days considering my next actions. I
may go to Bristol or Plymouth and look for jobs with my new resume
addition. Driving licenses are darn expensive to get! I have one now
and I shall look after it.
|Recently, I have seriously been considering moving all of my operations
to Bristol. There are far more opportunities up there. Though I have
enjoyed my time in Kernow.
|I have moved all downloads off site to the DSL connection, this allows
me to reduce the hosting charges for this website but does decrease
reliability. Most users of this web site are here to read the price
list and see what services I offer, not to download archived source
code, so this shouldn't be much of a problem to most customers.
|The power went off today. I think I have restored all services.
Powercut must have lasted about an hour.
|Carmack and Harmer have been reset by a thunderstorm. This affected the
OverChat service. Downtime == 40 minutes. Sorry for any inconvenience.
|Attention programmers, I have had no signal on my cell phone since Saturday.
I have no idea why, either. So if you can't contact me, you know why.
|The new LAN is complete. Both the Jeramy and I are using the internet
at the same time and we have ordered an ADSL connection. It feels good
to be back at my desktop at long last and I shall keep this log updated
with our developments as usual. It seems the banner on my site is not
working correctly at this time, this will be investigated.
|Hopefully I'll be moving next Thursday, Pater is booking up a man and van.
That'll take a few hours to get there! Certainly slower than the train.
It should be good to actually see everywhere along the route from the van
though. I had better start packing up all my CDs and records in the morrow.
Jeramy's brother will be there next week when I go along, so it should be
good, the Jeramy says he is going to cook some _hot_ food. I have ordered
a 120 GB disk for the core server on the new LAN.
|I am back, researching material for a project Jeramy and I are working on.
There will be a few problems here and there until we're working on one
server with version control. Oh, good news, I read that C99 has
snprintf( ) which has been needed for a long time and will make future C
code a lot safer. I'll have to grep through all Daybo Logic's code
looking for sprintf() uses. Unfortunately there is no easy, safe
alternative unless the compiler supports C99.
|What a great environment Open Watcom C/C++ is to work with. Nice to get
the old nerdy feel back. Different editor, debugger, compiler etc.
None of this integration nonsense. Apart from it saving DOS style CRLF
sequences at the moment I'm finding it perfectly adequate. I'll see if
there's an option for dealing with that issue. If I don't deal with it
the UNIX compilers hate the code. I'm off to Cornwall for a week on a
jobhunt. Hopefully making it a permanent residence as soon as possible.
|I resigned from the Newport committee of the Isle of Wight Conservative
Association in preparation for my move to Penryn, Cornwall. I have been
the secretary of the committee for a bit over a couple of years. I am
trying to order 1000 feet of cat 5e cable at the moment and having
difficulty because of billing Vs shipping addresses.
|I have updated the plan.cgi so that I can now add to this plan from the
site. I intend to allow other users to use it too, so they can use it as
a simple tool to document their activities. They think blogs are a new
phenomenom you know? Obviously they never heard of UNIX .plan files.
|Daybo Logic's router's dialler script has been replaced by a Perl script
which chooses which ISP to dial randomly. The old internet.sh has
been replaced with a wrapper to call the Perl program.
|I have written a simple PNG viewer CGI program for this site which will
help out, as this server seems to constantly make people downlond PNG
files because it's ignorant as to what they are, this Perl program will
put the right MIME type up so they are viewable. lsw32.exe has been
modified to call pngview.cgi on all PNG files. By the way, I have
compressed all the legacy binary CGI code with UPX. That should save me a
little space, although the cw3230.dll wouldn't compress. Also, I am not
going to be uploading ancient versions of libraries and things any longer,
it's a waste of time, also .tar.gz files will no longer be uploaded. Use
bzip2! I have changed policies to signing MD5SUM.txt files rather than
signing each program in a directory. If you want .tar.gz versions of
files you will have to mirror them yourself I have modified my software
section to make it more presentable. By the way, I am looking for a job!
|I have decided to write a Perl compatible Split() / Join() for C using
rather unconventional C programming, I am planning to return a vector of
strings which the caller will have to pass to a special clearing function
when they're finished with the split strings, dlstrlib_SplitClean( ) most
likely. Peter and Norma are coming over today from Scotland.
|I've been working on a lot of cool CGI stuff lately. Random banners,
picture viewers for our sister site and the latest cool tool is an SMS
cgi. When you're actually browsing the site I get an SMS alert to tell
me who's on the site. Pretty neat eh? Oh and apparently LoopyLoo has
been reading my .plan, I sees j00 LoopyLoo =P
|I have lent Stallman to my grandmother. I don't expect I'll see it again
for about 18 months or something.
|Our dynamic hostname finally expired. To be honest there's not much point
in me running services now anyway. It might was well wait until I have
broadband, which will be in a few years or something! =\
I have deleted ~/work/game/rttb/ as this really isn't the primary tree for
rttb any longer. The current one is reliant on the OverChat tree and all
I need is a simple way to build it when it's not in the tree before the
next RTTB release. OverChat's tree has to have the cutting edge stuff
|I have changed the normal/urgent notification to first and second class
mail, this will help people understand what it actually means. It doesn't
affect the actual email priority which is part of the header. All those
are normal unless somebody knows my current direct dynamic address and
sends email without the use of the form.
|There is now a new way to contact Daybo Logic, when using my
email page, if "urgent" is selected, the email is
available from my cell phone and I get an SMS alert that it has
arrived. This is highly useful for global availabillity and links the
email and SMS world. Linking email with SMS is always a spam risk but
since the underlying email address is not public, it shouldn't become a
problem. If it ever does, the service will of course be withdrawn.
|Daybo Logic have discovered that an item of our mail was
opened. Recently, we sent some Daddy's sauce to somebody in the USA. It
was opened and inspected before it got there, well, that's to be
expected. However, a close business partner of Daybo Logic sent us a
package in the post the other day and it arrived with a black piece of
duct tape on it (to close it) which wasn't there when it was sent. It
would seem even internal mail is being inspected in this country. What's
going on here? I'd appreciate any stories you have like this, so I know
it's not me spcifically!
|I have not got around to the DNS updater, I must do that, as I must do
loads of things, tapes to record etc. I am restoring a backup to restore
the source code of lsw32 (a component of this website). Then I must send
it back to Auntie Nicky's place. It has finally stopped raining and I am
going to deliver some leaflets for Daybo Logic. I have applied for a job
down at a security place. Also, if I don't get that, I have my beady eyes
on another job which will be from 05:00 to 09:00 every day. Strange hours
for a strange person!
|The DNS for ddrp.daybologic.com is down again. Poxy thing, well, Stallman
messed up and the networking wouldn't work, so I installed NT 4 and I am
sending that machine to my grandmother for a while. I'll set up an
appropriate DNS updater on Stallman shortly. Loads of new adverts printed
which I am going to deliver soon. Hell, maybe this business will start
making some money one day. Thanks to gooch for the recent daybodep patch,
I'm going to release the new version of daybodep today.
|Sorry the DNS for ddrp.daybologic.com has been down recently. The usual
machine which handles this (Stallman) is off, I'm using it's monitor for
Haas. Stallman is a Win95 machine, I shall move the DNS updating program
to Carmack within the next 24 hours and restore access.
|I have replaced the hub with a switch, store and forward, blocking,
128KB per port I think. Uplinked as before, network diagram is slightly
wrong because Katz is connected to the main switch still. I have updated
the internal DNS record for ddrp.daybologic.com to point to Silverman so
that ddrp.daybologic.com resolves whether we are connected to the internet
or not, before this, daybo-logic.homeunix.net could not be resolved when
we were not online. Happy new year. Current plan is that OverChat will
support chat on March 5th 2003. I am planning a wireless network for the
street if I can find some inexpensive equipment. This plan will be
moving to my website to support viewing when I am not online and to make
|Fault: Hub 100Mbps circuit has failed. Will be purchasing a new switch
soon, in the mean time Haas and Harmer are offline.
|This is a tribute to Samantha Wille who died this morning, aged 25. It's
no age to die and she we be sorely missed. I had known her for quite a
while, not as long as her other friends though. She was beautiful and
thoughtful. I shall miss her saying "ch00n after ch00n after ch00n after
ch00n". Suicide is a selfish act, if you're reading this and thinking
about it, don't forget, you may not be able to handle your pain but do you
really want to hurt your friends?
|Bored with working on OverChat for so long, decided to have a quick break
and work on a ZIP cracking program. Also, today the NTP service was
restored, turned out that the Daybo Logic stratum 3 server config file has
been overwritten by a clock-drift file somehow.
|ddrp.daybologic.com is back, via daybo-logic.homeunix.net, fed up with all
the broken UNIX implementations of the dynamic DNS client, I have switched
to a Windows version, this is is no way an endorsement of Windows.
It's more the case that www.dyndns.org are especially pedantic
and a lot of clients fail miserably to go by their rules.
|Trying to put the strings in OverChat as a whole into external RTTB files,
this is neccersary to support Russian.
Got an enquiry about some networking some chap wants to do. I think I'll
go down the supermarket and enquire about advertising and then go to the
|Last night I lost my .gaimrc file due to gaim loading before the share to
Torvalds was established, then quitting after it had been established.
Writing a new empty .gaimrc over my proper one. In future I'll
back .gaimrc up to .gaimrc~ every now and then. At the moment I have
cracked open the backup from 15th September 2002.
|OverChat has had some changes to make sure it doesn't annoy people by not
letting them sign on again if they became disconnected. Planning a
registration by the client feature but it's causing some trouble.
Progress has been slow but I intend for this to improve shortly. Dagsylad
joined the effort by making another client (codenamed Sarah). Just
purchased another hard disk for Torvalds, 20GB. Although actually, I am
half tempted to turn my main workstation into a FreeBSD boxen. Currently
my BIOS doesn't like the 10GB drive it's got in it and won't allow more
than 2GB to be used. Perhaps the 20GB is the answer? ie. move this 10GB
into Torvalds and put the new 20GB in Carmack?
|About to release OverChat 0.1.1, lots of changes, mostly boring or
security. I know people want chats but it just can't be done at the
moment, it wouldn't be sensible, there are far more important things
to do, such as error return codes from the server... but I'll be doing
it soon, even if it's slow & buggy.
|My previous utility I mentioned in my previous plan update is not complete
yet but I do intend to complete it. In any case I am fixing my
authentication scheme in OverChat at the moment, ready for the next
release of OverChat (0.1.0), which is behind schedule. I have been on a
promotion/advertising short course in order to aid OverChat and a
computer help setup. Daybo Logic has decided to transform the site from
just a development site into a computing help site. You can still find all
the old stuff here and it will all be kept updated when I change my code
but the main focus will be business. This is a .co.uk after all :-)
|ddclient doesn't seem to always want to update my IP, so I'm using ddup
now, hopefully things will improve. I expect a new version of OverChat
very soon now.
|OverChat is now capable of selecting an encryption algorithm from
Oberchatten's popup menu in conversation windows, users can select a
passphrase here too. As long as the key matches the other end of the
conversation, it will work, Oberchatten can automatically detect the
algorithm the message is encoded with.
|I have finally fixed the DNS issue, so ddrp.daybologic.com is back online.
ipcheck.py was messed up and had caused my account to be marked abusive,
so I have recreated an account by the same name to fix it, I am now using
ddclient, which stays resident as a daemon. During discussions on
irc.digiserv.net #chat, we made ourselves a perfect authentication scheme
for OverChat. I'm not going over it again though. First year of plumbing
course is over. OverChat is now using ciphered text and I am creating a
library of cryptographic routines for use by OverChat.
|Poll results came in with the majority of users wanting existing bugs fixed,
such as error codes (messaging a user who is not online for example).
Whois also rated highly. 0.1.0 of the server will be released to fix all
known bugs. I have installed a MySQL server on harmer.daybologic.com,
this will help me migrate the OverChat user database to a distributed
model. Due to a bug in previous Oberchatten clients, 0.3.5 cannot work with
the new server, users are required to upgrade to Oberchatten 0.4.0 as soon
as it is released. I've caught up with my plumbing now, just a bathroom
suite to fit and some homework to do and hopefully I'll get through. Ian
Johnson thinks I should do the NVQ 3 plumbing (advanced stuff), I don't
know. Installed Python because my IP updating script wasn't working (no
Python interpreter installed), apologies to anybody who was trying to
fetch my plan. FTP server is down too. Upgraded to FreeBSD 4.6 and rebooted
after 97 days uptime on Silverman. Total Konfuzion has made me a second
|OverChat 0.0.19 has been released, I don't think Haesu has got it running
yet, this is really the first server which doesn't have a writable socket
picking problem. The other night, 6 of us were on my 56k modem connected
test server, no problems, no lost messages. I've sent a poll to the
mailing list to see what feature they want the most for the next version.
It should conclude on Sunday. I'll be busy soon. I have a lot of work to
do for the next month to try to complete my plumbing course. There's a
great deal of risk I won't catch up. Happy Overchatting.
|I have put up a new map of the Daybo Logic network.
|I got the issues with OverChat fixed and made a release not so long ago,
my current job is to fix two exploits in the system and add killabillity
to operators, that's nearly complete but it does require a bit of thought,
such as notifying the person who has been killed or sending an error packet
back to the killer and error codes haven't been used in Blake yet so they
don't work. These will need to be implemented for messaging somebody who
is not online too.
|After a brief meeting with Haesu I have decided to deal with the NetBuffer
issue and have a release out by the weekend. Wish me luck. SafeAllocator
will be suspended and will be worked on when nothing more important with
OverChat is going on.
|I am currently working on a new protected system for memory management
within OverChat (Blake) so that I can get TK and SBeast onboard without
them accidentally messing up any memory. Whilst slower than normal memory
management this new method will ensure OverChat dies straight away with
stats should any memory be overwritten.
|I have decided to start logging what's going on around here on this site
again. Many things have changed. We now use a 200Mbps full duplex switch
from Netgear (FS-105). We switched our domain to daybologic.com internally
rather than ddrp.lan, all our IPs changed and we joined the TowardEX VPN.
Which allows other organisations on the VPN to access all our internal boxen
directly, even though we're behind NAT. Our DNS for daybologic.com and
overchat.net. overchat.net is our new site which is dedicated to OverChat,
our new chat system.
|Today has been a good day, integration of NetBuffer into OverChat went quite
well and the new autoconfigurator is safe.
|Seem to get the occasional PPP mess up, have to send it the KILL signal
and then 'route delete default' before restarting
|I have been aware that ddrp.daybologic.com has had a major DNS outage for
at least two weeks, I think I may have found the problem. The Python
script or crond is no longer working and needs a fix, it will be ready
|Critical! The other day, on a very hot summer day, haas.ddrp.lan
died. What this means for me is I now have no Windows boxes to compile
things on. What this means to everybody else is that
http://ddrp.daybologic.com (the webserver running Sambar) is no more, at
least for now. I think haas.ddrp.lan burnt, the motherboard fried or
something, I smelt burning before is just died. It beeped for a while but
then it was gone.
|The index page on the Daybo Logic site locally (Isle of
Wight) disappeared, strange. Luckily I recovered it from the site hosted
|It seems the domain daybologic.com is up and working again. All hail
|There seems to be a DNS outage, ns1-hl.twdx.net answers DNS requests about
daybologic.com and for some reason it isn't. Commercial interest: TowardEx
owns ns1-hl.twdx.net and has kindly donated this service to Daybo
Logic. Use the backup dynamic DNS name
|The G4 (torvalds) froze up again. I think it's caused by running out of
swap space, 'tis what I suspect anyway. Takes ages to get the thing
running again with such a big /usr partition to scan. Linux torvalds
2.2.17-0.6.1 #1 Thu Jun 29 15:15:56 CDT 2000 ppc unknown
|Another BIND outage. Turns out I am not using BIND 9 at all! Even so, I'll
have to get a script working to make sure it's automaticlly restarted
|Critical! torvalds.ddrp.lan has hung, uptime only 9 days! I was eventually
logged out of my telnet terminal. It started with the mouse disappearing
on torvalds's console, eventually the machine wouldn't even respond to
pinging. Whoa! The mouse optical light has gone out, this is a severe
hardware freeze up! Power cycled torvalds!
|The page on BTinternet's proxy came back! I think ppp is still dialing the
old number, not sure why the page was gone temporarily, anyway, killed ppp as
firstname.lastname@example.org and restarted ppp -auto -nat, been a while since
I've had to do that. The modem made different noises when it redialled, I
guess we're on another number, sorted.
23:00 BIND 9 died again. As per usual, I have no idea how long this lasted
until I spotted it.
|ISP BTInternet are supplying a single page on
their proxy! I could try to find a public, anonymous proxy
but they are going to get rid of the phone number anyway. Luckily, to
avoid using their download and self config rubbish I clicked on the Mac
config and it says the phone number. Updated
(email@example.com:/etc/ppp/ppp.conf) with the new phone number. The
page has gone.
|Out of power sockets, listend to the radio and had to unplug the modem,
bought a new power strip for non-critical things like the radio so all's
|Tried to clear the proxy cache and logs, Squid didn't like it. In my
ignorance I rebuilt all the source for it rather than re-create the cache
with ./squid -z. log and cache directories are now owned by nobody.root,
not sure if this is the way it was before but it's all working.
|BIND 9 seems to have core dumped again, most services would have been
unavailable all night! Grr. I need a new BIND or at least some sort of
script to automatically start named.
|Critical! Power outage for less than 5 seconds on ring main torvalds (up
81 days) goes down. Older machines reboot automatically. Rewired the
network, 11 sockets now used, re-ordered UTP cables in IP order. The
network was down for 6 hours because I was out, switched all power off. It
took 30 minutes or so to restore things in this order: Boot gateway
(silverman), boot torvalds, restore squid cache, port forwards, 15 mins
taken by disk scans and the like. Everythings seems to work again now,
except trying to use SWAT from haas.
|DNS outage as BIND 9 dies again, quietly on silverman.ddrp.lan (or
10.0.0.1 should I say). Restarted as firstname.lastname@example.org, I don't know
how long the outage was for but the web and FTP access would not have
|Gateway problems seem gone now, load is nothing generally. FTP problems,
two people have reported PORT commands not working, I don't know what's
|The web server is hosted on an NT box (haas.ddrp.lan) and decided it would
warn me about buffers not written all day long, anyway, had to reboto in
the end and I don't think the web server was affected (but for the
|The constant traffic is keeping load on the gateway 0.80 - 0.98 ish Maybe
FreeBSD can't take it? It's an old machine, P166 with 16MB RAM, might have
to find some more RAM, not too much though.
|Disconnected (normal, happens at 2 hour intervals). The gateway didn't
redial. Telnetted into it to try to restart ppp, odd things happening with
the default route, just did a quick and dirty silverman# reboot as root,
blah. Set new port forwards 80->haas:80, 21->torvalds:21
8023->torvalds:23. Restarted ppp. Went to bed. The gateway redialed
within 10 mins.
|Wiping a file pushed load on the proxy up to 5.00, FTP server will have
become sluggish and it was really hairy trying to surf the web in here!
|The gateway rebooted! I can't see what's going on either
seeing as it has no monitor. I was justing su-ing root on it and it locked
up and rebooted. Only took 10 mins to get reconnected to the internet
|Been up and down all day, dynamic DNS wasn't working well, had to do it
manually. In the end I set the Python script up in a cron job every five
minutes. Means email@example.com will get a ton of mail though when
the link is down. Anyway, if any program accesses the internet while I am
out, within 10 mins the FTP server should be accessable to all!
|Daybo Logic's gateway's IP address is now known all the time! I have
installed Python on the gateway, along with ipcheck.py which everytime
/etc/ppp/ppp.linkup is executed updates my IP. My hostname is
|I can't tell when because of my super-caching proxy but the Daybo Logic
FTP mirror supplied by Total Konfuzion has been restored. I have been in
touch with it's administrator and he told me the problems with the strange
content were due to a mistake he made while setting up new FTP server
|At this time the new DNS is working OK, however, the CNAME from
www.daybologic.com to www.daybologic.co.uk does not work, something to do
with the fact that donald.fasthosts.co.uk is a shared server, virtual
hosting or whatever. Also today the HTTP and FTP services were restored
with port forwards through the gateway.
|Most of today has been taken up by installing the new gateway,
ddrp.daybologic.com will have been down all this time but is now
back. With a FreeBSD gateway with automatic dialling ppp when anybody
tries to route packets to the internet. When the network got back online
it used the new alias, ddrp.dnsalias.org
|daybologic.com nameservers changed from Easyspace to TowardEx
Technologies, Easyspace wanted to bill me for the DNS service but, they
owe me £40 so they can lump it. I will probally change the registrar for
daybologic.com too. Expect daybologic.com to be down for a while now! Use
my UK version
|I received this from CyberID (www.daybologic.co.uk's webhost):
** ESSENTIAL MAINTENANCE **Our technicians will be carrying out essential
maintenance on Fileserver 2 on the NT4 system. This will occur at around
midnight on Sunday the 18th March. It is estimated that this work will
take approximately 6 hours. We apologise for any inconvenience that this
may cause. As always, the outage will be kept to an absolute minimum.
|Dynamic DNS problems on overlord.hldns.com. HLDNS is causing me trouble,
their Windows based client doesn't want to update the DNS unless I tell
it to and even then it doesn't always work. They don't even have a UNIX
based updater. I think the time has come to switch my DNS services to
DynDNS and see how I get along with them. If you've been using the proper
CNAME, ddrp.daybologic.com this will hopefully make no difference to you.
|The Daybo Logic FTP server provided by Total Konfuzion appears to be
suffering from downtime, I have not been in contact with Total Konfuzion
to know what is happening.
|British Telecom and their multiple nameservers, one was down which my
nameserver was forwarding requests to so my network was effectively down
for a short while this morning until I got hold of another IP for forward
to and reset the nameserver. ns1.ddrp.lan. Network down this afternoon,
playing with silverman.ddrp.lan to see if I could dial up through FreeBSD
so I could use this as a firewall.