[886] in Moira
Re: MIT.MIT.EDU: no update_server
daemon@ATHENA.MIT.EDU (dkk@MIT.EDU)
Tue May 16 15:04:49 1995
From: dkk@MIT.EDU
Date: Tue, 16 May 95 15:04:40 -0400
To: tom@MIT.EDU
Cc: postmaster@MIT.EDU, carla@MIT.EDU
Cc: moiradev@MIT.EDU, moira-admin@MIT.EDU
In-Reply-To: <9505161654.AA01984@fahrvergnugen.MIT.EDU> (message from Tom Coppeto on Tue, 16 May 95 12:54:33 BST)
> We should put on the moira queue to trigger an error when
> update_server is not running..
The current failure model appears to allow for two conditions:
soft fail -- keep retrying, and don't make a big deal of it
hard fail -- don't retry until reset, and make a big deal of it
Either we change that model (moderate software change), make this a
hard failure (small software change), or adapt our operational
procedures to catch this sort of thing more quickly (no software
change, but possibly a script to be written). I'm for the operational
change, if practical... (The dcm did log a warning, and that did show
up as one line in the Daily Gazette.)
If there's enough support for a different solution (eg: updating the
dcm program), I'll go along with it, but looking at the source code
(/mit/moiradev/src/update/client.c) I don't see a solution I like.
