URGENT --- SNMP v3 issues ---

[cico_lt]

Hi,

I have a strange behaviour with my Zabbix 3.0.1.

If I run an "SNMP get" command from the shell, I receive data, but with a simple Item in Zabbix,
it returns data some time and the other times it doesn't return anything.

It write "item xxx became not supported: Cannot connect to "xxx.xxx.xxx.xxx": Authentication failure (incorrect password, community or key).

After some variable time, the Item return "supported" ... only for one or two check. Then it returns "unsupported".

I have only ICMP and that Item on that device and ICMP Items aren't in "unsupported" state.

Furthermore, that Item is only on that device.

Is it the same issue like https://support.zabbix.com/browse/ZBX-7007 ???

Thanks in advance

[batchenr]

Hi,

I have a strange behaviour with my Zabbix 3.0.1.

If I run an "SNMP get" command from the shell, I receive data, but with a simple Item in Zabbix,
it returns data some time and the other times it doesn't return anything.

It write "item xxx became not supported: Cannot connect to "xxx.xxx.xxx.xxx": Authentication failure (incorrect password, community or key).

After some variable time, the Item return "supported" ... only for one or two check. Then it returns "unsupported".

I have only ICMP and that Item on that device and ICMP Items aren't in "unsupported" state.

Furthermore, that Item is only on that device.

Is it the same issue like https://support.zabbix.com/browse/ZBX-7007 ???

Thanks in advance

see if other then this item if you have unsupported item on this host
and disable them all.
see if it helps - maybe create a little script that run this snmpget command with logs so when it becomes unsupportd youll see the last data it gets.

[DukeNukem3D]

Same behavior for different hosts, version 3.0.8

[DukeNukem3D]

Strange thing - I changed StartPollers to 1 and 'Authentication failure' messages has gone away, now graphs are looking good. Do not know what is wrong with SNMPv3 processing..

[eertul]

Strange thing - I changed StartPollers to 1 and 'Authentication failure' messages has gone away, now graphs are looking good. Do not know what is wrong with SNMPv3 processing..

Hi all, I had the same issue and, I changed "StartPollers" to 1 on Proxy, it started to get data.
What is the point of this?

Thanks.

[enkikun]

Hi. Same problem for me. Just turned "StartPollers" to 1. All seems start to work so far.

[DukeNukem3D]

Hi all, I had the same issue and, I changed "StartPollers" to 1 on Proxy, it started to get data.
What is the point of this?

Thanks.

It seems I have found the root cause. After hours of searching internets, I have found mention about duplicate engine IDs in SNMPv3 and that it causes NET-SNMP to behave this way - it 'thinks' that data SNMPv3 response is spoofed since engine IDs the same while source IP addresses are different. Setting "StartPollers" to 1 resolves this since Zabbix starts using 1 thread of SNMP poller rather than several threads in parallel, so at each given time only one device is being polled. In my case problem was in several Cisco ASA devices configured in multiple VDC mode active/active cluster - bot active and standby nodes in each VDC are represented with the same engine ID. Solution was not elegeant - I stopped monitoring standby nodes using SNMP, since I did not find how to change engine ID on standby node. Anyway, upon failover, active address moves to standby node and monitoring is still consistent.

You can check for duplicate engine IDs using this OID 1.3.6.1.6.3.10.2.1.1 - simply get it's value using script from all of your devices and check for duplicates.

[enkikun]

You can check for duplicate engine IDs using this OID 1.3.6.1.6.3.10.2.1.1

There is no duplications for me:
[root@zabbix ~]# snmpwalk -Ofn -v3 <creds_for_host1> 1.3.6.1.6.3.10.2.1.1
.1.3.6.1.6.3.10.2.1.1.0 = Hex-STRING: 80 00 00 09 03 00 00 21 D8 CA 54 C0
[root@zabbix ~]# snmpwalk -Ofn -v3 <creds_for_host2> 1.3.6.1.6.3.10.2.1.1
.1.3.6.1.6.3.10.2.1.1.0 = Hex-STRING: 80 00 00 09 03 00 00 15 2C CB 82 00

But i have same issue.