Hello Nathan,
I have been having a look at the postfix pmda and the memcache pmda to get a sense of how to go about making a fix
I think I need a bit more help.
If I understand this correctly, the steps are;
1. new function named call_qshape to call qshape for a given queue, read the data and update $caches{$qname}
2. set up file pmda timer (one for each queue) to run the above function
The above two steps wil replace postfix_fetch_callback and postfix_do_refresh.
Am I on the right track ?
Also I can't see the use of the 'cluster' variable in postfix_fetch_callback. It does not get passed in
Thanks
On Fri, 2016-08-12 at 14:49 +1000, Chandana De Silva wrote:
I think that makes sense. Let me see
On Thu, 2016-08-11 at 23:29 -0400, Nathan Scott wrote:
I would punt that the qshape command is blocking for at least one of the queues,
for a long time. Best way to resolve this is probably in postfix_do_refresh() -
if this was run on an interval timer (see pmdamemcache.pl for example), instead
of serially during postfix_fetch_callback(), the problem should be resolved.