On 07/18/2011 03:49 PM, André Schild wrote:
Am 18.07.2011 15:38, schrieb Raffael Bachmann:
Thanks for the fast answer.
How many workers would you suggest?
I think this depends on the number of clients which are actively
accessing your sogo server.... ;)
There are about 30 users with thunderbird open all day. And they all
subscribed many other calendars.
And to have a more sophisticated parent process which "loops once
trough" all children
and if none is found sleep for 0.1 seconds and then retrys.
And if in the second try still all children are busy, do log this as a
warning in the logfile...
I think a system where the worker reports after finishing a request
might be better.
The parent process would only need to check every now and then to see if
some really strange things are happening, like crashed workers or
workers busy for a long time with one request.
Raffael
--
users@sogo.nu
https://inverse.ca/sogo/lists
