Hi Everyone,
We're using SOGo 1.3.11 here for some thousands of users, and we can say
it works well.
Thanks for that !
But sometimes we have some problems with some sogod process : indeed
sometimes they take 100% of CPU during a 'long' time (> 1min.).
That can appear when IMAP is down, or when a caldav client sends some
specific requests ...
... Actually we're not sure to know why this problem appears exactly at
the moment.
SOGo (sogod parent - the WatchDog) kills the sogod children that don't
respond after 10 minutes, that's the "watchdog request timeout".
So if a problem appears on some child process of SOGo when they handle a
request, after 10 minutes they are killed and we are sure that SOGo
becomes again available.
That's nice :-)
But we thought that 10 minutes is too much, so we changed it.
Indeed, we found in the Defaults.plist file (which provides many others
advanced parameters) the parameter WOWatchDogRequestTimeout, which is
fixed to 10.
We set up this WOWatchDogRequestTimeout to 1 in our .GNUstepDefaults and
it seems to work :-)
But is it safe to put 1 for this parameter ?
Indeed, why the default value is "so high" (10 minutes) ?
Also, it seems that we can't setup this parameter lower than 1 minute,
like 30 seconds for example, why ?
Thanks for your response,
Vincent Bonamy.
--
[email protected]
https://inverse.ca/sogo/lists
