On Wed, Apr 02, 2014 at 04:28:48AM +0530, KK Patnaik wrote:
> 1) Please find the output of qshape in the link http://pastebin.com/pQ4h4pNY
Sorry, I don't read pastebins. You need to post the output (or at
least lines for the top 20 or so domains) to the list.
> 2) Please find the "top" output at the time of peak hours
>
> top - 13:22:52 up 10:22, 3 users, load average: 5.36, 5.62, 5.62
> Tasks: 276 total, 1 running, 275 sleeping, 0 stopped, 0 zombie
> Cpu0 : 1.0%us, 0.4%sy, 0.5%ni, 90.9%id, 7.2%wa, 0.0%hi, 0.0%si,
> 0.0%st
> Cpu1 : 1.9%us, 1.6%sy, 0.5%ni, 71.0%id, 24.3%wa, 0.1%hi, 0.7%si,
> 0.0%st
> Cpu2 : 0.9%us, 0.9%sy, 0.3%ni, 92.4%id, 5.3%wa, 0.0%hi, 0.3%si,
> 0.0%st
> Cpu3 : 0.9%us, 1.6%sy, 0.5%ni, 71.0%id, 25.1%wa, 0.2%hi, 0.8%si,
> 0.0%st
> Mem: 4138404k total, 457076k used, 3681328k free, 44916k buffers
> Swap: 4096532k total, 8k used, 4096524k free, 151240k cached
System is idle, which suggests high network latency, but impossible
to tell without analyzing the logs!
> 3) I ran the collate script sent by you. Checking that.
It will show the collated history of various messages in the queue,
based on the qshape output's top domains, look for any clues to
the delay experienced by the corresponding domains. If these
destinations are throttled in the queue manager look for the
messages for which delivery is actually attempted:
postfix/smtp[pid]: qid: to=..., ... dsn=...
rather than immediately deferred because the destination is throttled:
postfix/qmgr[pid]: <qid>: to=...
> Can you please suggest on the basis of qshape output and the CPU load?
Once you post the qshape output and some collated log entries for
representative messages for highly delayed destinations. Feel free
to obfuscate only the local part (user name) of the sender and
recipient addresses, but NOT the domains, the actual domains may
be essential to understanding why the queue is congested.
--
Viktor.
