Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards, Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.html or sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment.
Could you tell us more about your setup? How many devices are you monitoring? Whats the hardware on the Observium server? CPU, storage? What version of Observium, PHP and RRDtool are you running?
/Markus
Den mån 17 sep. 2018 kl 11:48 skrev Ming Wong, CLSA ming.wong@clsa.com:
Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards,
Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.html or sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Hi, probably this is already in progress ;)
Ming Wong, CLSA wrote on 17.09.18 12:35:
Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards,
Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.htmlor sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment.
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Hi Markus,
Thanks for the reply. Below are details the setup:
The Observium is on a VM running RH7. It has 4 Xeon vCPU, 8G memory, and 100GB local disk. 90 devices are being monitored, and the polling interval is 1 minute. RRDCache is enabled.
Software versions are Observium CE 18.9.9420, PHP 5.6.37, RRDtool 1.6.0, Mariadb 10.3.9 and RRDCache 1.6.0. The reason for running PHP 5 instead of PHP7 is to integrate with LDAP.
Below is the polling summary. SQL takes 32.47% of the time.
##### Completed polling run at 2018-09-17 16:01:36 ##### o Devices Polled 1 o Poller Time 53.21 secs o Memory usage 20.5MB (peak: 22.5MB) o MySQL Usage Insert[1/0.035s] Update[1243/10.366s] Delete[1/0.017s] Cell[6/0.062s] Row[1115/6.065s] Rows[119/0.683s] Col[4/0.051s](17.277s 32.47%) o RRDTool Usage update[1255/9.493s] (9.493s 17.84%) o SNMP Usage snmpget[13/3.125s] snmpwalk[16/8.485s] (11.61s 21.82%)
Thanks a lot, Ming
From: Ming Wong, CLSA Sent: Monday, September 17, 2018 5:36 PM To: 'observium@observium.org' observium@observium.org Subject: Observium poller is becoming slow
Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards, Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.html or sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment.
php7.0-ldap works fine with AD integration
Regards Christian
Från: observium observium-bounces@observium.org För Ming Wong, CLSA Skickat: den 17 september 2018 15:30 Till: 'observium@observium.org' observium@observium.org Ämne: Re: [Observium] Observium poller is becoming slow
Hi Markus,
Thanks for the reply. Below are details the setup:
The Observium is on a VM running RH7. It has 4 Xeon vCPU, 8G memory, and 100GB local disk. 90 devices are being monitored, and the polling interval is 1 minute. RRDCache is enabled.
Software versions are Observium CE 18.9.9420, PHP 5.6.37, RRDtool 1.6.0, Mariadb 10.3.9 and RRDCache 1.6.0. The reason for running PHP 5 instead of PHP7 is to integrate with LDAP.
Below is the polling summary. SQL takes 32.47% of the time.
##### Completed polling run at 2018-09-17 16:01:36 ##### o Devices Polled 1 o Poller Time 53.21 secs o Memory usage 20.5MB (peak: 22.5MB) o MySQL Usage Insert[1/0.035s] Update[1243/10.366s] Delete[1/0.017s] Cell[6/0.062s] Row[1115/6.065s] Rows[119/0.683s] Col[4/0.051s](17.277s 32.47%) o RRDTool Usage update[1255/9.493s] (9.493s 17.84%) o SNMP Usage snmpget[13/3.125s] snmpwalk[16/8.485s] (11.61s 21.82%)
Thanks a lot, Ming
From: Ming Wong, CLSA Sent: Monday, September 17, 2018 5:36 PM To: 'observium@observium.org' <observium@observium.orgmailto:observium@observium.org> Subject: Observium poller is becoming slow
Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards, Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.html or sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment.
Hi, As Christian said, you can just install php7-ldap to have LDAP-auth work with PHP7, Observium fully supports PHP7 for all features.
As you are running 1min-polling you will have a lot higher stress on your system (5x the stress to be exact ;) ) so you will need faster hardware to scale this up. I suspect that the current bottleneck is your storage, as this is a VM its probably backed by some SAN that is not that quick on IO. Do you monitor the Observium-host itself with Observium? If not, install snmpd on the machine and then add localhost in Observium. Wait like 1hour to gather some stats and then go to localhost>Graphs>Processor. The bottom graph there should be "Extended Processor Utilisation", please screenshot it and send us here.
/Markus
Den mån 17 sep. 2018 kl 15:59 skrev Christian Eriksson < christian.eriksson@advania.se>:
php7.0-ldap works fine with AD integration
Regards
Christian
*Från:* observium observium-bounces@observium.org * För *Ming Wong, CLSA *Skickat:* den 17 september 2018 15:30 *Till:* 'observium@observium.org' observium@observium.org *Ämne:* Re: [Observium] Observium poller is becoming slow
Hi Markus,
Thanks for the reply. Below are details the setup:
The Observium is on a VM running RH7. It has 4 Xeon vCPU, 8G memory, and 100GB local disk. 90 devices are being monitored, and the polling interval is 1 minute. RRDCache is enabled.
Software versions are Observium CE 18.9.9420, PHP 5.6.37, RRDtool 1.6.0, Mariadb 10.3.9 and RRDCache 1.6.0. The reason for running PHP 5 instead of PHP7 is to integrate with LDAP.
Below is the polling summary. SQL takes 32.47% of the time.
##### Completed polling run at 2018-09-17 16:01:36 #####
o Devices Polled 1
o Poller Time 53.21 secs
o Memory usage 20.5MB (peak: 22.5MB)
o MySQL Usage Insert[1/0.035s] Update[1243/10.366s] Delete[1/0.017s] Cell[6/0.062s] Row[1115/6.065s] Rows[119/0.683s]
Col[4/0.051s](17.277s 32.47%)o RRDTool Usage update[1255/9.493s] (9.493s 17.84%)
o SNMP Usage snmpget[13/3.125s] snmpwalk[16/8.485s] (11.61s 21.82%)
Thanks a lot,
Ming
*From:* Ming Wong, CLSA *Sent:* Monday, September 17, 2018 5:36 PM *To:* 'observium@observium.org' observium@observium.org *Subject:* Observium poller is becoming slow
Hi,
The polling time for each device is becoming long when more and more devices added to Observium. From the poller debug output most of the time is spent on SQL update. For a network switch, each port will trigger a SQL update.
Is it possible for Observium to poll all ports first, and then have a single SQL update statement covering all ports ? Any advice in enhancing the polling speed or database performance ?
Thanks and regards,
Ming
The content of this communication is intended for the recipient and is subject to CLSA Legal and Regulatory Notices. These can be viewed at https://www.clsa.com/disclaimer.html or sent to you upon request. Please consider before printing. CLSA is ISO14001 certified and committed to reducing its impact on the environment. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
participants (4)
-
Christian Eriksson -
Markus Klock -
Mike Stupalov -
Ming Wong, CLSA