NOTE: This is an archived copy of the Dragonfire Internet Services website. Dragonfire is no longer in operation; please do not attempt to contact any of the addresses listed on this site.

Dragonfire System Event Log

Note: All times on this page are given in Universal Coordinated Time (UTC), also known as Greenwich Mean Time (GMT).

Current system status

No known problems.

Please note that URLs will no longer function; we found it necessary to disable this method of FTP file access to improve server performance. You should instead use your hostname in FTP URLs; for example:

Often, problems accessing Dragonfire are the result of Internet backbone troubles. The following network service providers have pages listing the status of their networks. We would appreciate links to any additional network status pages.

Historical system events (most recent first)

Note that while we try to cover all system and service outages, some problems do not leave any traces of their existence if they resolve themselves before we detect or are notified of them. Also note that, as a rule, outages which pertain to a single service only and which are less than ten minutes in duration are not listed here.

30 November 1998
Server Bahamut rebooted to fix a kernel bug preventing access to some sites. Server downtime: 1.5 minutes.

30 November 1998
Counters are restored.

27 November 1998, 13:40 UTC
Scheduled maintenance; in order to complete the maintenance, services are disabled for some time. Total downtime: 4 hours (approximately).

23 November 1998, 02:24 UTC
Server palantir rebooted. Server downtime: 2 minutes.

22 November 1998, 04:05 UTC
Mail service is temporarily disabled. Due to a system misconfiguration, a certain message was being repeatedly sent from the system to itself, replicating in the process and taking up large amounts of disk space and CPU time. The problem turned out to be easily solvable, but in order to restore the system to normal it was necessary to empty the mail queue. This may have resulted in a small number of normal messages being lost as well; we apologize for this. If you have not received a message you were expecting, you may want to ask the sender to resend it. Service downtime: 15 minutes.

18 November 1998, 23:34 UTC
Server palantir rebooted. Server downtime: 1 minute.

13 November 1998, 15:11 UTC
Server palantir rebooted. Server downtime: 10 minutes.

11 November 1998, 06:54 UTC
Both servers are taken out by a power failure, and for unknown reasons, do not return when the power is restored. Total downtime: 2 hours, 14 minutes.

8 November 1998, 20:47 UTC
Both servers rebooted. Total downtime: 5 minutes.

5 November 1998, 18:35 UTC
Server palantir noticed to be down and rebooted. Server downtime: unknown.

1 November 1998, 17:04 UTC
Server palantir rebooted. Server downtime: 15 minutes.

30 October 1998, 21:25 UTC
DNS and HTTP servers restarted. Service downtime: 10 minutes.

28 October 1998, 00:26 UTC
Server palantir stops responding. Rebooting the server does not solve the problem, but after some investigation, it is determined that server Bahamut was "eating" packets intended for palantir. A reboot of Bahamut solves the problem. Server Bahamut downtime: 11 minutes; server palantir downtime: 1 hour, 33 minutes.
24 October 1998, 14:44 UTC
Server palantir rebooted to fix weirdness with the FTP daemon. Server downtime: 1 minute.

23 October 1998, 13:26 UTC
A routing problem apparently with prevents some access to Dragonfire for about 15 minutes.

21 October 1998, 15:00 UTC
Some virtual servers were down for approximately 3 hours as the result of a bug in the monitoring program, which has since been fixed.

17 October 1998
Mail delivery service suspended temporarily for minor software configuration changes. No mail was lost as a result of this maintenance.

16-17 October 1998
Due to server modifications for moving all users to basic domain service, HTTP service for various sites was down intermittently.

7 October 1998, 20:00 UTC
Both servers reboot. Total downtime: 30 minutes. (Some virtual servers remain down for an additional 20-25 minutes.)

3 October 1998, 22:47 UTC
Server palantir rebooted. Server downtime: 2 minutes.

26 September 1998, 01:06 UTC
Server palantir rebooted. Server downtime: 1 minute.

24 September 1998, 01:24 UTC
HTTP daemon upgraded to a new version. HTTP service downtime: 5 minutes.

23 September 1998, 22:05 UTC
Server Bahamut reboots. Server downtime: 11 minutes.

22 September 1998, 10:13 UTC
Filesystem corruption is discovered on server palantir, and all services on that machine are taken offline to repair the damage. No user files are believed to have been affected, although this damage did affect the ability for certain users to create directories or use CGI scripts. Server downtime: 10 minutes.

18 September 1998, 08:21 UTC
Server palantir rebooted. Server downtime: 1 minute.

17 September 1998, 02:18 UTC
Perl binary replaced on server palantir due to reports of failing CGIs which led to the discovery the binary was apparently corrupted.

15 September 1998, 03:11 UTC
Server palantir rebooted. Server downtime: 1 minute.

10 September 1998, 04:21 UTC
Some virtual servers were down for approximately 25 minutes. The cause of this is unknown.

30 August 1998, 05:05 UTC
Server palantir is rebooted when FTP service begins acting up. Server downtime: 1 minute.

25 August 1998, 13:43 UTC
Server Bahamut (which had been experiencing no difficulties) was mistakenly rebooted by a CAIS technician. Server downtime: 11 minutes.

25 August 1998, 07:23 UTC
Server palantir goes down and needs to be rebooted. Server downtime: 6 hours, 27 minutes.

21 August 1998, 22:54 UTC
Server palantir rebooted when FTP slows to a crawl. Server downtime: 3 minutes.

18-19 August 1998
Logs indicate that some virtual domains on server Bahamut were inaccessible for a total of 8 hours, from 11:00 UTC to 15:00 UTC each day. The cause of this is unknown.

18 August 1998, 08:53 UTC
Server palantir rebooted when it is discovered that the HTTP servers on that machine are refusing to start, possibly due to memory corruption. Server downtime: 2 minutes, with some prior (unknown) downtime for HTTP servers on that machine.

13 August 1998, 17:18 UTC
Server Bahamut rebooted as the simplest solution after a runaway server process causes many other system processes to die. Server downtime: 3 minutes.

10 August 1998, 17:14 UTC
Server Bahamut is rebooted due to system overload. Server downtime: 13 minutes.

9 August 1998, 10:00-17:00 UTC
Logs indicate that some virtual domains on server Bahamut were inaccessible for 7 hours. The cause of this is unknown.

5 August 1998, 22:00 UTC
Server palantir is restarted after going down for an unknown reason. Estimated server downtime: 16 hours.

5 August 1998, 17:10 UTC
Server Bahamut rebooted due to an unknown system error. The system may have been inaccessible for a short time before, extending the actual downtime. Server downtime: 14 minutes.

4 August 1998, 17:37 UTC
Both servers are rebooted. System activity indicates that HTTP servers may have overloaded the system. Server downtime: 6 minutes.

25 July 1998, 17:49 UTC
Server palantir rebooted due to NFS-related problems. Server downtime: 3 minutes.

15 July 1998, 17:48 UTC
Server palantir crashes, and is rebooted. System activity suggests that the problem may be NFS-related, and we are therefore working on reducing NFS traffic between the servers. Server downtime: 11 minutes.

11 July 1998, 08:20 UTC
Server palantir crashes again, and is rebooted again. Server downtime: 4 hours, 45 minutes.

9 July 1998, 13:44 UTC
Server palantir crashes, is rebooted, and minor filesystem damage is fixed. Server downtime: 1 hour, 26 minutes.

6 July 1998, 16:47 UTC
Server palantir is rebooted as part of a test of new system monitoring software. Server downtime: 1 minute.

4 - 5 July 1998
Server palantir fails, unfortunately during a holiday break when the servers are being watched as closely. The server is discovered disabled (not actually down, but nearly so) at 22:48 UTC on 5 July 1998 and is rebooted and repaired. Server downtime: 33 hours (estimated from reports).

30 June - 1 July 1998
A misconfigured mail delivery program is not taken into account during system cleanup, causing mail delivery to some Dragonfire administrative addresses to fail. User accounts were not affected by this error.

26 June 1998, 12:23 UTC
An HTTP daemon serving a number of virtual servers is seen to be down and is restarted. Prior downtime is unknown, since the system monitoring tool had not reported any problems. The tool is reconfigured to look harder for and yell more loudly about problems.

24 June 1998, 02:50 UTC
Server palantir fails. Inexplicably, a hard reset does not start it up again, but it recovers after being power-cycled. Server downtime: 29 minutes.

23 June 1998, 18:51 UTC
Server Bahamut begins experiencing the same problems experienced earlier in the day by server palantir, and is immediately reset to stave off any additional service problems. Server downtime: 13 minutes.

23 June 1998, 09:10 UTC
An unknown HTTP server problem causes the main server,, as well as several virtual servers, to stop responding. Eventually this is traced to what appears to be a rare system bug occurring on server palantir, causing the init process (the "overseer" of the entire system) to freeze and leave many server processes unable to function properly. The system is given a hard reset, followed by correction of the filesystem corruption thereby caused. No user files are believed to have been damaged by this incident. Service downtime: 6 hours (approximately).

Note that we are looking into establishing virtual servers for all users to limit the damage that can be caused if the server processes fail again.

18 June 1998, 19:31 UTC
Server Bahamut rebooted due to odd error conditions appearing. Server downtime: 1.5 minutes.

14 June 1998
A bizarre NFS problem, cause still unknown, takes out a few virtual servers for most of the day before it is corrected.

8 June 1998, 15:21 UTC
Servers taken down for replacement of the UPS which had apparently not been doing its job. Total downtime: 47 minutes.

8 June 1998, 11:23 UTC
Servers reboot. (Another power flicker?) Total downtime: 10 minutes.

5 June 1998, 19:44 UTC
Servers hit by a power flicker. Total downtime: 10 minutes.

2 June 1998, 14:30 UTC
Servers hit by another backup power test at CAIS. Total downtime: 10 minutes.

1 June 1998, 18:33 UTC
Server palantir reboots. Server downtime: 2 minutes.

31 May 1998, 22:37 UTC
Server palantir rebooted because it was acting strangely, though FTP and web service was still functioning. Server downtime: 2 minutes.

30 May 1998, 06:46 UTC
Server Bahamut goes down. Attempts to restart it appear to be successful, but only for a short time; eventually it is determined that the power supply is faulty and needs to be replaced. We theorize that this failure is due to voltage spikes resulting from the multiple occurrences of power loss in CAIS's weekly power tests. However, whatever caused this failure, we were unfortunately not prepared to deal with it, and hence it takes a good deal of time to restore the server to normal operation. Server downtime: 16 hours, 17 minutes.

29 May 1998, 08:41 UTC
Server Bahamut goes down and needs to be manually rebooted. Server downtime: 2 hours, 20 minutes.

27 May 1998, 00:02 UTC
11 virtual servers are downed for approximately one hour to move them from one server to the other.

26 May 1998, 14:30 UTC
Both servers are downed again by a CAIS power test. A call to CAIS's NOC reveals that an electrician is scheduled to move VMA's power to CAIS's backup power system before next week's test. Total downtime: 10 minutes.

25 May 1998, 03:51 UTC
Both servers reboot. Total downtime: 10 minutes.

19 May 1998, 14:30 UTC
Servers are downed due to another backup generator test at CAIS. Total downtime: 10 minutes. A configuration error keeps one web account inaccessible for the next eight hours; this error has since been corrected.

12 May 1998, 14:27 UTC
A backup generator test at CAIS brings down both servers. A CAIS technician says that CAIS will be requiring all colocation customers to use CAIS's backup generator system, which should bring an end to these episodes of downtime. Total downtime: 15 minutes.

5 May 1998, 13:44 UTC
Another power failure strikes, bringing both servers down. Discussion with a technician at CAIS, the provider of the physical network link and colocation space, reveals a startling fact: Verio Mid-Atlantic (the company formerly known as ClarkNet, which provides the logical network connection) is not purchasing backup UPS power from them. According to VMA, they have their own backup power supply at CAIS, but it failed. Total downtime: 2 hours, 41 minutes.

27 April 1998, 08:21 UTC
Server palantir also needs to be rebooted because of a disk error. Server downtime: 13 minutes.

27 April 1998, 07:32 UTC
Server Bahamut needs to be rebooted because of a disk error (possibly a result of the power outage). Server downtime: 22 minutes.

27 April 1998, 05:27 UTC
Both servers are brought down by a power failure. Total downtime: 1 hour, 13 minutes.

23 April 1998, 08:35 UTC
Server Bahamut crashes again. The timing and frequency of these crashes, especially in light of the prior stability of the server (it had been up for over 40 days before the April 7 self-reboot), suggest to us that the server is under some sort of attack. ICMP packets (i.e. support for pings and traceroutes) are filtered out to see if that helps. Server downtime: 4 hours, 55 minutes.

23 April 1998, 01:21 UTC
Server Bahamut rebooted after kernel warnings pop up in the system log. Server downtime: 2.5 minutes.

20 April 1998, 13:22 UTC
Server Bahamut reboots; the root drive is not clean after the reboot and has to be checked manually. Server downtime: 59 minutes.

18 April 1998, 05:54 UTC
Server Bahamut bites the bucket and gets rebooted again. A new kernel (the latest in the "development" series of Linux kernels) is compiled to test on that server. The new kernel fails spectacularly due to conflicts with old software; the software problems are resolved, but the kernel itself then fails. The original kernel is restored. Total downtime: 14 hours (approximately).

17 April 1998, ~22:00 UTC
Both servers stop responding at about the same time for unknown reasons, and have to be reset manually. Total downtime: 7 hours (approximately).

14 April 1998, 14:14 UTC
Server palantir rebooted to fix an odd problem preventing a web server process which had failed earlier in the day from restarting. Server downtime: 1 minute. (The two sites served by that particular process were also down for several hours prior to the reboot.)

7 April 1998
Server Bahamut reboots. Server downtime: 1 minute.

28 March 1998
Server palantir taken down for installation of a replacement hard drive. Server downtime: 70 minutes.

26 March 1998
Several people report that they cannot reach the web server. We can find no local cause for this; all processes are running normally and there are no unusual entries in the system log. This suggests that the reported problems were caused by either Internet connectivity failures or a multiple service failure on Dragonfire (e.g. logging facility, HTTP daemon monitor, and HTTP daemon). System monitor scripts are fine-tuned to prevent (or at least greatly reduce the possibility of) such failures occurring again.

24 March 1998, 03:29 UTC
Server palantir rebooted. Server downtime: 6 minutes.

21 March 1998, 15:17 UTC
Server palantir crashes, and has to be rebooted. Server downtime: 32 minutes.

8 March 1998, 04:03 UTC
Server palantir rebooted. Server downtime: 2 minutes. Indications are that one or two HTTP servers may have been down for some time previously.

6 March 1998, 08:43 UTC
Server palantir rebooted. Server downtime: 2 minutes.

1 March 1998, 18:29 UTC
Server palantir rebooted due to apparent memory corruption. Server downtime: 1 minute.

25 February 1998, 16:11 UTC
Text/graphics counter (/cgi-bin/mcounter) recompiled and reinstalled, as the text counter seems to be working well.

25 February 1998, 05:17 UTC
Text counter (/cgi-bin/tcounter) code reviewed, recompiled, and experimentally reinstalled.

22 February 1998, 20:34 UTC
Server palantir reboots. Server downtime: 1.5 minutes.

22 February 1998
Intermittent (and possibly extended) service outages occur through about half of the day due to a disk-full condition on server Bahamut. Estimated total web/POP service downtime: 6 hours.

21 February 1998, 20:01 UTC
One HTTP daemon on server palantir killed and restarted because it had apparently stopped responding to requests.

17 February 1998, 07:45 UTC
Server palantir rebooted when FTP sessions suddenly start dying for no apparent reason. Server downtime: 1.5 minutes (FTP outage extends approximately 10 minutes farther back).

12 February 1998, 02:09 UTC
Server palantir reboots. Server downtime: 1.5 minutes.

5 February 1998, 00:25 UTC
Server palantir stops responding, and needs to be rebooted. The cause of the freeze is unknown. An unexpected result caused by file-sharing between the servers causes HTTP daemons on the other server (Bahamut) to stop responding. Total downtime: 80 minutes.

3 February 1998, 16:00 UTC
Corrected configuration problem denying access to anonymous FTP files through, and disallowed anonymous FTP to main server files through (virtual servers on the same machine are still accessible).

3 February 1998, 04:43 UTC virtual server (which serves Dragonfire logos) misconfiguration fixed after it was reported down.

1 February 1998, 23:08 UTC
Server palantir rebooted. Server downtime: 13 minutes.

31 January 1998, 00:59 UTC
A new server,, is brought online.

30 January 1998, 17:01 UTC
Main server taken down for a short time during installation of a new server. Total downtime: 15 minutes (approximately).

22 January 1998, 21:30 UTC
Re-enabling the FTP server caused the load to shoot back up again. At this point, the problem appears more likely to be excesesive disk thrashing from many FTP processes trying to access different files at once alongside the HTTP servers. Anonymous FTP has been disabled until we find a way to reduce the thrashing.

22 January 1998, 20:01 UTC
All FTP processes terminated again as the load problem continues. After process termination, the load drops quickly back to normal, suggesting either a problem in the FTP service or an FTP-based denial of service attack. FTP server downtime: 43 minutes.

22 January 1998, 17:25 UTC
All FTP processes are terminated due to extreme amounts of swapping on the server; the anonymous FTP user limit is then reduced from 200 to 100 users to prevent a repeat of the problem.

21-22 January 1998
Delays in mail delivery introduced by system load are found and corrected. Note that no mail was lost; however, some messages may have been delayed one or more days.

20 January 1998
ClarkNet provides a new set of IP addresses which bypass the ACSI backbone. Servers are reconfigured to listen to the new IP addresses as of 21 January, 01:10 UTC.

15 January 1998, 05:00 UTC
Counters disabled again after the HTTP servers grind to a halt.

15 January 1998, 04:41 UTC
Counters re-enabled to stop the flood of complaints from people who hadn't bothered to read this page.

10 January 1998, 05:17 UTC
Web counter temporarily disabled upon discovery of many "junk" counter entries and unusually large numbers of calls to the counter.

9 January 1998, 00:15 UTC
FTP/POP server mysteriously dies; since it has never failed before, the outage is not noticed until much later. POP service downtime: 17 hours, 25 minutes (approximately).

2 January 1998, 15:28 UTC
Server rebooted with a newer kernel to see if that stops whatever is slowing down the server. Server downtime: 20 minutes.

1 January 1998, ~16:30 UTC and continuing
Denial of service attack? All server processes on Dragonfire slow down; the delays cannot be traced to any program running on the server itself, suggesting some sort of external cause.

1 January 1998, 02:15 UTC
A bug is fixed in the HTTP server software, apparently eliminating the earlier problems with freezes.

29 December 1997, 19:24 UTC
A configuration file for the mail server is mysteriously corrupted, temporarily suspending receipt of mail. No mail is lost as a result of this error.

24 December 1997, 18:50 UTC
Web server upgraded to a new release. Web server downtime: 15 minutes, plus occasional restarts to fix configuration or installation problems over the next several hours.

24 December 1997, 16:28 UTC
After some intermittent service, the virtual domain server problem seems to have been resolved, and virtual domains are once again available as they should be.

24 December 1997, 14:21 UTC
The virtual domain server crashed inexplicably, and has not been able to be restarted. We are actively working to resolve this problem.

23 December 1997
Nameservice was reported to be down for a large part of the morning. We have not been able to identify any system faults during that period; we can only guess that this stems from a problem with our network access provider, which also provides primary nameservice for the domain.

10 December 1997, 00:47 UTC
Server taken down to replace a defective SCSI cable. Total downtime: 14 minutes.

9 December 1997, 21:34 UTC
Most services temporarily suspended to fix a filesystem error. Total downtime: 10 minutes.

9 December 1997, 20:45 UTC
Server reboots. Total downtime: 4 minutes.

29 November 1997, 13:11 UTC
Server reboots. Total downtime: 12 minutes.

26 November 1997, 14:54 UTC virtual server brought back online manually (again), this time fixing the problem for good.

22 November 1997, 02:56 UTC
Server crashes and has to be brought back online manually. We are investigating this and the previous crash as a possible denial-of-service attack against Dragonfire, though we are not hopeful of gaining much information after the fact. Total downtime: 53 minutes.

20 November 1997, 17:28 UTC virtual server (which serves Dragonfire logos) brought back online manually after it failed to come up normally during restart.

20 November 1997, 12:26 UTC
Server crashes, and does not restart as normal. Total downtime: 1 hour, 37 minutes.

19 November 1997, 15:53 UTC
Various configuration parameters on the HTTP server are modified to increase server performance. Web server downtime: 2 minutes.

18 November 1997, 20:07 UTC
DNS change reversed as it seemed to have adverse effects.

18 November 1997, 11:55 UTC
Traffic to split among four IP addresses to balance HTTP server load.

7 November 1997, 01:48 UTC
Server reboots. Total downtime: 12 minutes.

25 October 1997
An incorrect HTTP daemon is started up after system boot, and a few domain sites are unavailable for most of the day.

24 October 1997, 10:45 UTC
ClarkNet, during maintenance on their equipment, knocks a cable loose from Dragonfire which effectively brings the system down. Total downtime: 15 hours, 54 minutes.

23 October 1997, 04:07 UTC
System crashes due to disk drive errors (on the same disk which we tried to replace on multiple previous occasions but could not due to bad replacement drives). The drive is replaced and data transferred. Total downtime: 13 hours, 47 minutes.

18 October 1997
HTTP daemon restored to standard operation. With the exception of the fact that the server must still be restarted on occasion due to an apparent memory leak, the problems of the past month appear to be resolved.

9 October 1997, 16:00 UTC
HTTP daemon is replaced with a newer version which seems to function considerably better.

27 September 1997, 13:51 UTC
Server is taken down for scheduled maintenance to install a new 9GB disk drive. Total downtime: 3 hours, 45 minutes.

14 September 1997, 19:10 UTC
The primary HTTP daemon is taken down momentarily to test the effectiveness of an experimental updated version. Web server downtime: 5 minutes.

14-15 September 1997
Dragonfire's HTTP daemon begins to show signs of stress, restarting itself several times an hour.

11 September 1997, 03:47 UTC
Server reboots. Total downtime: 1 minute. A bug in the HTTP server causes wrong pages to be served for an additional 20-25 minutes before it is found and corrected.

4 September 1997, 03:36 UTC
Server reboots. Total downtime: 1.5 minutes; the HTTP daemon takes a bit longer to recover.

17 August 1997, 22:20 UTC
A strong thunderstorm takes out power to Dragonfire momentarily. Total downtime: 10 minutes.

15 August 1997, 04:14 UTC
Following unspecified maintenance work by ClarkNet, one of Dragonfire's hard disks was found to be damaged. Analysis and temporary repairs took most of the day. Total downtime: 21 hours, 22 minutes.

31 July 1997, 14:25 UTC
Dragonfire is shut down (intentionally, this time) to allow ClarkNet to replace the defective UPS. Total downtime: 2.5 minutes.

31 July 1997, 12:55 UTC
The UPS (Uninterruptible Power Supply) provided by ClarkNet for Dragonfire and several other machines serviced by ClarkNet overloads, taking Dragonfire down with it. Total downtime: 27 minutes.

5 July 1997, 13:18 UTC
Server is shut down for an upgrade to 128MB RAM and installation of a new tape drive. Total downtime: 43 minutes.

3 July 1997, 21:40 UTC
Dragonfire's HTTP daemon crashes and does not restart as usual. Web server downtime: 4 hours, 44 minutes.

1 July 1997, 17:15 UTC
Server is rebooted to cure a small but persistent resource leak. Total downtime: 1.5 minutes.

1 June 1997, 10:27 UTC
Server reboots. Total downtime: 9 minutes.

27 May 1997, ~03:30 UTC
Due to problems caused by BBN (, one of the Internet backbone providers, Dragonfire is cut off from at least a large part of North America. Normal routing is restored around 17:30 UTC.

16 May 1997, 05:53 UTC
Due to an error in configuration while switching IP addresses, the server becomes unreachable and must be power-cycled. Total downtime: 1 hour, 23 minutes. The problem is exacerbated by unannounced ClarkNet network maintenance, causing continued occasional network dropouts for a while.

3 May 1997, 23:56 UTC
Server reboots. Total downtime: 46 minutes.

24 April 1997, 02:48 UTC
During a network upgrade, ClarkNet changes the IP address of the router to which Dragonfire is connected, thus cutting Dragonfire off from the Internet, and fails to inform us of the change until considerably later. Total downtime: 11 hours, 47 minutes.

23 April 1997, 20:40 UTC
Server goes down due to a loose cable. Total downtime: 37 minutes.

20 April 1997
We receive a report that AGIS's network, through which most of Dragonfire's Internet traffic travels, is under attack. Due to this, many people may well encounter delays accessing Dragonfire.

20 April 1997, 16:23 UTC
Server rebooted to get rid of a runaway process. Total downtime: 1 minute.

19 April 1997, 20:52 UTC
Server reboots. Total downtime: 9 minutes.

31 March 1997, 19:04 UTC
Server crashes, and reboots automatically. Total downtime: 43 minutes.

28 March 1997, 18:07 UTC
Server is moved to another network center. Total downtime: 1 hour, 39 minutes (plus DNS propogation time).

27 March 1997, 05:21 UTC
Server crashes. Total downtime: 11 hours, 16 minutes.

21 March 1997, 17:27 UTC
Server is moved to a new network center. Total downtime: 6 hours, 46 minutes (plus DNS propogation time).

15 March 1997, 13:04 UTC
Server is taken down shortly after schedule to move drives over to a new case. Total downtime: 1 hour, 40 minutes.

11 March 1997, 00:05 UTC
Server is taken down again due to (yet more) problems with the drive. All data on the drive is moved to other filesystems, and the drive is removed. Total downtime: 1 hour, 38 minutes.

10 March 1997, 08:00 UTC
Server is taken down for scheduled emergency maintenance to repair the drive. The drive is fixed with little loss. Total downtime: 2 hours.

9 March 1997, 18:04 UTC
The troubled drive fails again, and is temporarily taken offline to prevent it from bringing the rest of the system down. Total downtime: 1 hour, 5 minutes.

8 March 1997, 02:10 UTC
One of Dragonfire's drives experiences a transient failure (likely from heat caused by the malfunctioning fan). Total downtime: 30 minutes.

8 March 1997, 00:55 UTC
New fan removed due to bad performance. Total downtime: 18 minutes.

7 March 1997, 14:33 UTC
Server taken down momentarily to adjust new fan. Total downtime: 1.5 minutes.

7 March 1997, 12:11 UTC
Server taken down to install a new cooling fan. Total downtime: 18 minutes.

26 February 1997, 10:05 UTC
Server rebooted to test a new and (hopefully) improved kernel. Total downtime: 4 minutes.

22 February 1997, 10:58 UTC
HTTP daemon died unexpectedly and did not return. We were not made aware of the problem before the following day and thus could not fix it at the time. The server itself crashed about a day later. Total downtime: 11 hours, 40 minutes; web server downtime: 36 hours, 39 minutes.

21 February 1997, 00:39 UTC
Server crashed for unknown reasons and rebooted automatically. Total downtime: 10 minutes.

19 February 1997, 17:25 UTC
Kernel was modified and server rebooted to attempt to fix the recent crashes. Total downtime: 1.5 minutes.

19 February 1997, 15:01 UTC
Server crashed for unknown reasons. Total downtime: 1 hour, 1 minute.

19 February 1997, 07:36 UTC
HTTP daemon crashed for unknown reasons, and was restarted manually as soon as the problem was noticed. Total downtime: 3 hours, 30 minutes.

19 February 1997, 03:27 UTC
Server crashed unexpectedly; stable kernel was reinstalled. Total downtime: 28 minutes.

16 February 1997, 10:24 UTC
Server was rebooted multiple times to test added system functionality. Total downtime: 20 minutes.

15 February 1997, 16:22 UTC
Server was rebooted to install a new kernel version. Total downtime: 4 minutes.

9 February 1997, 14:48 UTC
Server was shut down for minor rearrangement of drives to improve air flow. Total downtime: 5 minutes.