Philip Hazel wrote:
>
> On Mon, 28 Apr 1997, William Craven wrote:
>
> > Exim 1.61 was launched on the main MailServer on Tuesday 15th April.
>
> > Further investigation lead to the wait-smtp
> > DB being corrupted - dumpdb, fixdb & tidydb all failed with a
> > segmentation fault !!
>
> > Has anyone experienced this problem and if so how did they rectify it?
>
> Which DB library are you using? There have been cases of corruption
> reported in the past using Berkeley DB 1.85. Some fixes were put into
> Exim 1.60, and there was one other fix in 1.61:
We are running Exim 1.61 using Berkeley DB 1.85 - I had to double check
the loader to make sure it did load the correct library which it did.
Please note Exim is running on Solaris 2.5.1.
> We haven't had any such problems here on our Solaris 2.5 systems using
> ndbm, but then we probably don't have such a heavy load (the busiest
> yesterday made 46,000 deliveries).
Could it be that Berkeley DB cannot handle DBs of a certain size or more
? The reason I ask is that I have implemented a daily trimdb run to
trim the DBs that have entries more than 7 days old. At the time of the
problem Exim was running for over 10 days and the DBs were not being
trimmed and the wait-smtp DB was very large (cannot recall the size).
If it occurs again I'll make sure that I save the core and the DB.
Wm.
--
William Craven Email: William.Craven@???
University Computing Services Tel: +1-604-822-8955
University of British Columbia FAX: +1-604-822-5116
Vancouver, BC, Canada V6T 1Z2
