Gaps in SNMP graph

[TheWitness]

Get a bigger box, if it's a VM takes < 5 minutes and not even a reboot. I would say have at least 12 cores. 24 is better.

[kribbas]

Yes agreed but until then I need to get these hosts back up. Right now the its staking less than 5 min and working fine except around 20 snmp hosts that are stated as down, however if I do snmpwalk they respond properply. Any ideas on how to get them back up? Since not all are broken, only a few. I have tried adding again but does not seem to help.

[kribbas]

Could it help to restart the snmp daemon or are there any risks to that?

[TheWitness]

No, and best bet is to reduce the MAX OIDS for those devices. Many less powerful switches and routers can not do that many in on get request. Like I said, if you don't know what it does, go slowly until you understand.

[kribbas]

So you mean that they host show as "down" because they time out, and the time out is a result of too high MAX OID per get request? (10)?

Do I change the MAX OID in the poller settings or in the device settings?

[TheWitness]

Device. And yes

[TheWitness]

Code: Select all

UPDATE host SET max_oids = 5;

[kribbas]

Ok,

Im down from 10 --> 4 and still the device is shown as "DOWN" in the device list, when getting device info it does not say "SNMP ERROR" at the top but show the device info (as if it was not down)...

[kribbas]

Regarding:

CODE: SELECT ALL
UPDATE host SET max_oids = 5;

Do I need to update in mysql as well? Is it not enough to update in the device GUI settings?

[kribbas]

Im not getting any info in the log, since it says the hosts are down it seems the poller is ignoring these hosts even if I change the MAX OID.

[TheWitness]

Did you run that MySQL command? That does them all. If the host are being hit with those scripts, you might have to set script servers to 1 again. You never uploaded the script BTW.

[kribbas]

I did not run it, don't want to change them all, since it has been working fine before.

So if I change the script server to 5 in the GUI then I changed it back to 1 yesterday, do I need to revert somewhere else as well?

[kribbas]

The change in MAX OIDs does not seem to have an impact. The devices are stated as DOWN but I know that they are not, so something seems to be off with cacti fetching through SNMP?

[kribbas]

Is there any log file where I can monitor the SNMP attempts?

[TheWitness]

You can turn on selective device debug in the general settings for the device or edit device and then turn on debug from the upper right hand corner.