statistics_collector takes 12-15 hrs to complete

[gennolo]

I am running Plesk 12.5.30 on several Centos 6.7 VMs.
From some days some VMs takes hours to complete the nightly "statistics_collector" thing (up to 12/15 hours) and increases load average of the VMs themselves.

Usually the process had always took 3-4 hours to complete and I am puzzled in troubleshooting this.

"ps auxf" on the involved boxes show this situation for the most part of the process:

root      60556  0.0  0.0  19052   604 ?        Ss   03:01   0:00 /usr/sbin/anacron -s
root      61928  0.0  0.0 106060   756 ?        SN   03:26   0:00  \_ /bin/bash /usr/bin/run-parts /etc/cron.daily
root      61952  0.0  0.0 106060   744 ?        SN   03:26   0:00      \_ /bin/sh /etc/cron.daily/50plesk-daily
root      61954  0.0  0.2 355788  4208 ?        SN   03:26   0:00      |   \_ /usr/bin/sw-engine -c /usr/local/psa/admin/conf/php.ini /usr/local/psa/admin/plib/DailyMaintainance/script.php
root      63590  0.0  0.2 354504  4208 ?        SN   03:27   0:00      |       \_ /usr/bin/sw-engine -c /usr/local/psa/admin/conf/php.ini /usr/local/psa/admin/plib/DailyMaintainance/task-script.php ExecuteStatistics --period=daily
root      63595  0.0  0.2 359640  4208 ?        SN   03:27   0:00      |           \_ /usr/bin/sw-engine -c /usr/local/psa/admin/conf/php.ini /usr/local/psa/admin/sbin/statistics --no-webstat
root      63600  0.7  3.7 153452 71452 ?        DN   03:27   2:34      |               \_ /usr/local/psa/admin/bin/statistics_collector
root      61953  0.0  0.0 105948   608 ?        SN   03:26   0:00      \_ awk -v progname=/etc/cron.daily/50plesk-daily progname {?????   print progname ":\n"?????   progname="";????       }????       { print; }

So it appears statistics_collector staying in "uninterruptible sleep" status.

BTW - What is this process doing ?

awk -v progname=/etc/cron.daily/50plesk-daily progname {?????   print progname ":\n"?????   progname="";????       }????       { print; }

I had problems with statistics_collector in the past resulting my VMs to have high I/O (I upgraded disks to more performant hybrid-NFS storage and the problem seemed partially resolved until today).

Any idea to better troubleshoot what's happening or what part on statistics take so long to complete?
Note that I have already disabled web-statistics on all domains.

Thanks for the help.

 

[IgorG]

Duplicate thread http://talk.plesk.com/threads/best-way-to-completely-get-rid-of-statistics_collection.335925 ?

 

[gennolo]

Hi Igor,
My mistake - I had the previous topic bookmarked to the "old" Odin forum (http://forum.odin.com/) and I didn't realized it was migrated here too so I started a new discussion.
Feel free to close / delete old thread or to move this update in the old discussion.