Hey all,
I’ve got a device (Zhone MXK-F) with over 10,000 ports on it, and Observium is taking well over 5 minutes to poll it (upwards of 15 minutes, actually). While there isn’t a definition specifically for MXK-F, it’s getting detected as MXK. os.inc.php is setting max-rep 50, so it’s at least trying to bulk walk, but I don’t know if it’s actually working. Overriding to max-rep 100 doesn’t have an affect on the polling time.
So, I guess my question is whether or not I’m on the right track in assuming that this issue is likely that of bulk walk on this device, or possibly some other issue, and how would I determine one way or the other? If it’s something that I can pinpoint, I can throw it back to the vendor to fix.
From the digging I’ve done so far, all other poller processes called from poller-wrapper complete in under 5 minutes, so I’m fairly certain that this issue is one of the device, not Observium. But, to be thorough, I moved my database and RRD storage over to local SSDs, and also enabled rrdcached - all that to ensure I didn’t have any I/O issues.
/pollerlog/view=devices indicates that it’s just the one device that has high timing (although other devices of the same type are spots 2-7 on the top 10 list high timers, but none is over 300 seconds except for this guy).
root@monitor:/home/jlixfeld# /opt/observium/poller.php -VV Observium 17.9.8837
##### Software versions #####
o OS Linux 3.16.0-4-amd64 [amd64] (Debian 8 (jessie)) o Apache 2.4.10 o PHP 5.6.30-0+deb8u1 (OPcache: DISABLED) o Python 2.7.9 o MySQL 5.5.55-0+deb8u1 (extension: mysqli 5.5.55) o SNMP NET-SNMP 5.7.2.1 o RRDtool 1.4.8 (rrdcached 1.4.8: unix:/opt/observium/rrdcached.sock)
##### Memory Limit #####
o PHP Unlimited
##### MySQL mode #####
o MySQL
##### Charset info #####
o PHP UTF-8 o MySQL utf8
##### Timezones info #####
o Date Monday, 02-Oct-17 12:55:15 EDT o PHP -04:00 o MySQL -04:00
root@monitor:/home/jlixfeld#
Thanks all!