[ZBX-3798] snmp host not being disabled even if not reachable Created: 2011 May 11 Updated: 2017 May 30 Resolved: 2011 Sep 19 |
|
Status: | Closed |
Project: | ZABBIX BUGS AND ISSUES |
Component/s: | None |
Affects Version/s: | None |
Fix Version/s: | None |
Type: | Incident report | Priority: | Critical |
Reporter: | Fco. Javier Rial | Assignee: | Unassigned |
Resolution: | Duplicate | Votes: | 0 |
Labels: | None | ||
Remaining Estimate: | Not Specified | ||
Time Spent: | Not Specified | ||
Original Estimate: | Not Specified | ||
Environment: |
centos 5.5 64 bits |
Attachments: |
![]() |
||||||||
Issue Links: |
|
Description |
I have a snmp host into zabbix, that is unreachable (I know, it's normal, it will be avaliable soon, but at this moment, port 161 it's filtered) : "Timeout connecting to 161" The problem is that zabbix_server keeps trying to get snmp values from it.. it nevers stop trying... this host never gets disabled by zabbix_server.. this items get stuck in the zabbix queue.. and it's affecting zabbix_server performance because the host has ~ 500 snmp v2 items.. Attached log file. As you can see in the log file, zabbix_server thinks "SNMP Host [SNMP-HOST-XXXX]: first network error, wait for 15 seconds" can be related with http://www.zabbix.com/forum/showthread.php?t=20876 ?? Thanks |
Comments |
Comment by Fco. Javier Rial [ 2011 May 11 ] |
Just to clarify: Host never gets disabled.. and zabbix keeps trying to monitor. |
Comment by Alexei Vladishev [ 2011 May 16 ] |
What version of Zabbix you are running? |
Comment by Fco. Javier Rial [ 2011 May 17 ] |
zabbix_server version 1.8.2.. maybe it's something fixed in new versions??? I checked out the release notes of the new versions but didn't find anything about such a bug like this. thank you |
Comment by dimir [ 2011 Jun 08 ] |
Can not reproduce with latest 1.8 . After disabling snmp host everything is handled properly: 3018:20110608:113916.177 SNMP Host [dimir]: first network error, wait for 15 seconds After that there is no any OID requests towards failed snmp host. Please, could you try the same with 1.8.5? |
Comment by Fco. Javier Rial [ 2011 Jun 08 ] |
I'll try with latest 1.8.x, but I'll need some time to upgrade because of customizations done in 1.8.2.. If you are sure that this issue is not affecting latest version in 1.8.x, please, feel free to close this issue, because I have seen in my installation, that in 1.8.2, snmp hosts get disabled when they are unrecheable.. I don't know what it's happening with this host. For now, I disabled it from monitoring, and when it becomes available, I will enable again. Thanks for your info and help. Cheers. |
Comment by richlv [ 2011 Jun 08 ] |
thanks for the speedy response, closing for now. please reopen if reproducible |
Comment by dimir [ 2011 Jun 08 ] |
Thanks. Actually, there are pretty big chances it's already fixed because in 1.8.2 the whole process of disabling/enabling a host was rather broken and there were a lot of changes since then in that matter. |
Comment by Aron Ujvari [ 2011 Aug 31 ] |
With Zabbix 1.8.5 this is issue is a problem. A powered down host correctly detected as unreachable (by active Zabbix Agent monitoring) but zabbix server poller processes are continously trying to get SNMPv3 values from it without success. This way all the pollers are delayed by SNMP timeouts as strace tracing is revealed. As there are hundreds of SNMP values to get from the unavailable hosts this causes massive delays in all the poller processes. Because of this are thousands of items in my zabbix queue with 1-5-5+ min. delay. On the GUI the Zabbix monitoring indicator is red, but SNMP monitoring is green. Monitoring using SNMP or Zabbix Agent can work with quite different timing pattern. It would be nice if there were dedicated pollers to collect data with SNMP and also it would be nice if different number of processes could be configured these pollers. With dedicated pollers for different types of monitoring these interferences caused by timeouts could be softened. |
Comment by richlv [ 2011 Sep 12 ] |
snmpv3 one might be |
Comment by dimir [ 2011 Sep 19 ] |
Reopen do resolve duplicate. |
Comment by dimir [ 2011 Sep 19 ] |
Let's move to the https://support.zabbix.com/browse/ZBX-4026 where the work has started. Closing duplicate. |