Random SNMP network error after new iDrac update

[g3ph4z]

Hello Guys!

First of all, my english is really bad, so sorry for my english mistakes.

I have 2 Dell servers, with iDrac 8. 3 weeks ago, I updated both to 2.52.52.52.
And after that, the Zabbix started to lost connection for few minutes.

log:

  6254:20180601:170611.182 SNMP agent item "NetStatus.[4]" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180601:170641.972 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6718:20180602:003901.117 SNMP agent item "MemoryEnum" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:003941.591 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6626:20180602:051611.871 SNMP agent item "GlobalSystemStatus" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:051641.750 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6806:20180602:121441.981 SNMP agent item "TempCritLowLimit.[2]" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:121511.022 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6302:20180602:202111.625 SNMP agent item "PowerUsageMinIdle" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:202211.769 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6849:20180602:202512.270 SNMP agent item "VoltageStatus.[5]" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:202541.324 resuming SNMP agent checks on host "THESERVERNAME": connection restored
  6653:20180602:231711.268 SNMP agent item "TempCritLowLimit.[1]" on host "THESERVERNAME" failed: first network error, wait for 15 seconds
  6849:20180602:231741.216 resuming SNMP agent checks on host "THESERVERNAME": connection restored

but when Zabbix alert me "NO IDRAC DATA > 5 MINUTES" I checked manually with SNMPWALK and worked fine.

Zabbix version: 3.4.9
OS: CentOS
DB: PostgreSQL

UPDATE:
I got call from Dell Support, and they told me, if we use snmpwalk, then I have to ask Zabbix support.....

UPDATE2:
I upgraded the snmpwalk FROM 5.7.2 to 5.7.3 from sourcecode it seems to be fine NOT fully, cause I still get "first network error" but it doesnt wait for 5-10 minutes and now I dont get any error.

What should I try?

[Colttt]

Hi,
try to turn off bulk snmp.

I also had found some iDRAC bugs in the past, for example if I try OID X and then OID y I got wrong values, I wrote a bug report to dell and after a while they fixed that.

[g3ph4z]

Hi,
try to turn off bulk snmp.

I also had found some iDRAC bugs in the past, for example if I try OID X and then OID y I got wrong values, I wrote a bug report to dell and after a while they fixed that.

Hi,

thank you for ideas, Now I turned off the bulk requests, I hope it will solve the issue.

[kloczek]

It is possible to reproduce those SNMP timeouts errors using plain net-snmp commands.
Problem has been many times discussed on zabbix forums and the issue with net-snmp code running on SNMP agent size.
You need to start chace Dell support. They need to start investigate those issues causes.

[g3ph4z]

It is possible to reproduce those SNMP timeouts errors using plain net-snmp commands.
Problem has been many times discussed on zabbix forums and the issue with net-snmp code running on SNMP agent size.
You need to start chace Dell support. They need to start investigate those issues causes.

I sent an email to Dell Support, I hope they can help us.
BUT
I called the our national Dell Support, and they told me I have to downgrade.

So, my last chance is the regional Dell Support Team.

Thank you for help.

UPDATE:
I got call from Dell Support, and they told me, if we use snmpwalk, then I have to ask Zabbix support.....

[Colttt]

when you stop the snmp queries (like disable the host) and wait a few minutes (eg 15-30min) and then enable it, does it work for the first minutes? if so did it again and run tcpdump until it happend, then stop it and take a look into it and what are the latest OID they try successfully and which one was the first htey didn't work. And then try this OID by yourself with snmpget, if that happen again it is a bug in the dell firmware

[g3ph4z]

Sorry for long delay. I will check this on the next week, and thank you

[kloczek]

You know .. slow reply of the SNMP agent you can reproduce even with running snmpd on localhost.
Really this issue has nothing to do with what is possible to observe on network layer.

[g3ph4z]

I guess, but I cant increase the timeout time? (and thank you too)

[glardz95]

hello,

do you resolve the issue, i have the same issue

[glardz95]

how do you update snmpwalk FROM 5.7.2 to 5.7.3 ?