July 20, 2004

Bio-Med mail web server

Below is a note from Dan regarding a problem with the Library's mail web server due to an over-grown log file. This has had an impact on users logging into Ovid for the first time - they would see an error message saying "Out of IDs for TC x500. This log file may also have affected sent e-mail and our web forms from Sunday, 7/18 to earlier today. Please note that lost e-mail cannot be recovered.

Jim

Hello,


It appears, from external phenomena that only came to light today,
that biomed10 (our main webserver) started having disk space issues on
its /var partition at approximately 1:00pm Sunday. This caused the
following problems:


1. Updates to Ovid's ID databases stopped. This caused 'out of IDs'
errors for first-time Ovid users (not a common commodity during the
summer). I am pursuing this problem with Ovid and expect to have it
fixed by EOD.
2. Outgoing mail loss. Any application that sends mail (such as
webmail forms, including document delivery and BIS research
requests) experienced mail loss during this period. Mails sent
during this period cannot be recovered.


The exact cause was a runaway log file, the 'ssl_engine_log', which
had grown to 1.3GB before it was caught. This log is unnecessary and
shouldn't have been enabled in the first place. I will turn off
writing to this file tonight.


I apologize for the interruption in service on these two fronts. I
have launched a short-term project to audit our webserver
configurations to make sure there aren't any other accidents waiting
to happen.


Regards,


Dan

Posted by biomedref at July 20, 2004 02:12 PM
Comments
Post a comment









Remember personal info?