Is using Berkeley DB 4.0.14 recommended?

[Aidan Evans]

On Sat, 19 Oct 2002 at 10:49 Rob Siemborski wrote to Aidan Evans and Info Cyrus

>On Sat, 19 Oct 2002, Aidan Evans wrote:
...
>>   My experience with DB 4.0.14 on RedHat Linux 7.2 is two instances of
>> complete failure of deliver.db and tls_sessions.db (and consequently all
>> Sieve processing) when the number of lockers reached 50,000, both instances
>> after about a week to ten days up time.  I expect a third failure if I let
>> the number of lockers get that high again.
>
>I don't think I've ever seen numbers that high, so I can't comment.

  Does CMU have a regular Cyrus shutdown/restart and/or regular reboot of
the machine running Cyrus?  Rebooting certainly resets the lock count; I
don't know if just a Cyrus restart is enough but I suspect not.

  We've unfortunately had a couple of unscheduled reboots with the side
effect that the lock count has not got too high since the last incident.

Aidan Evans   | Networks & Systems
(902)494-3332 | University Computing & Information Services
              | Dalhousie University, Halifax, N.S., Canada