[483] in Moira
Re: Mailhub ALARM CONDITION
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Sat Nov 28 00:47:35 1992
Date: Sat, 28 Nov 92 00:47:09 EST
From: tytso@Athena.MIT.EDU (Theodore Ts'o)
To: The mailhub slave checker <root@tsx-11.MIT.EDU>
Cc: network@MIT.EDU, bug-moira@Athena.MIT.EDU
In-Reply-To: The mailhub slave checker's message of Fri, 27 Nov 92 23:24:03 EST,
Well, it seems we've been a victem of several failure modes, not all of
which have been explained yet. The first is that the mailhub checker
hung on Wednesday, Thursday, and Friday morning. I have no idea what
went wrong, since it's worked before, and its derived from the Kerberos
slave checker, which hasn't given us any problems. When I tried running
it by hand, it worked and immediately sent out a problem report. I'll
have to watch it and see if it hangs tomorrow morning. In any case,
that's why we didn't get warned of this earlier.
Secondly, the dates on the aliases file indicates that athena managed to
get a successful alias update Wednesday night, but for some reason
athena-as-well did not:
athena has an out of date /usr/lib/aliases: Wed Nov 25 23:19:37 1992
athena has an out of date /usr/lib/aliases.dir: Wed Nov 25 23:59:52 1992
athena has an out of date /usr/lib/aliases.pag: Wed Nov 25 23:59:56 1992
athena-as-well has an out of date /usr/lib/aliases: Tue Nov 24 23:55:32 1992
athena-as-well has an out of date /usr/lib/aliases.dir: Wed Nov 25 00:02:16 1992
athena-as-well has an out of date /usr/lib/aliases.pag: Wed Nov 25 00:02:16 1992
I have no idea why the alias propagation failed; when I tried building
the Wednesday, 11/25 aliases file, the mailhub had no problem building
its DBM files based on it.
Worse yet, for whatever reason, Morira has _not_ generated a new aliases
file since Wednesday night:
----------------------------------------------------------
% mrcheck
Service ALIASES Interval 1410 Enabled/Idle/NoError
Generated Nov 25 23:05:07 1992; Last checked Nov 25 23:05:07 1992
Last modified by mar.root@ATHENA.MIT.EDU at 18-nov-1992 18:32:19 with mmoira
* Service has not been updated
Host ALIASES:ATHENA-AS-WELL.MIT.EDU Enabled/Failure/InProgress/Normal/NoError
Last try Nov 25 23:05:07 1992; Last success Nov 24 23:05:08 1992
Last modified by mar.root@ATHENA.MIT.EDU at 18-nov-1992 18:33:09 with mmoira
* Host has not been updated
Host ALIASES:ATHENA.MIT.EDU Enabled/Success/Idle/Normal/NoError
Last try Nov 25 23:05:07 1992; Last success Nov 25 23:05:07 1992
Last modified by mar.root@ATHENA.MIT.EDU at 17-nov-1992 13:19:22 with moira
* Host has not been updated
3 things have failed at this time
-------------------------------------------------------------------------
Note that for ALIASES:ATHENA-AS-WELL, the last success and last try
dates are different; yet no error was recorded, and the time something
was last modified was over a week ago. Strange.
Richard tried forcing the server to generate a new aliases file, by
first doing a reset error (which had absolutely no effect), and then a
reset service, which also wasn't enough to force moria to generate a new
aliases file. Because some dcm extracting was going on, he didn't want
to do further mucking about, and said he would look into it further from
home.
What I've done for now is installed the Wendesday night/Thursday morning
aliases file on athena-as-well, which installed without any problems.
Since there probably wasn't many changes or account registrations
happening over Thanksgiving, we can probably let this go one more day.
If by tomorrow things haven't gotten betta via-a-vis moira, I will try
calling Mark Roesnstein and ask him to look into things.
- Ted
