-
Problem report
-
Resolution: Fixed
-
Trivial
-
6.2.0
-
None
-
Sprint 92 (Sep 2022)
-
1.5
Summary:
I started to use the new MongoDB plugin.
Everything works fine for several hours, then after a while I get only "ZBX_NOTSUPPORTED: timeout occurred".
I have to restart the zabbix agent to get it working again.
My environment:
- Zabbix Agent 2 version: 6.2.0
- MongoDB plugin version: 1.0.0 (branch: master, commit: 21d652197aba9a54f4af45424a8d2ef98f5539b4)
(Faster) Reproduction:
It is possible to reproduce the bug in a few seconds/minutes by spamming the 'mongodb.server.status' requests:
while [ 1 ]; do zabbix_get -s 127.0.0.1 -p 10050 -k 'mongodb.server.status["tcp://localhost:27017","XXX","XXX"]' || break; done
Details:
Once the bug occurs once, it occurs for all subsequent requests (even when testing the URI of another mongodb server).
$ zabbix_get -s 127.0.0.1 -p 10050 -k 'mongodb.server.status["tcp://localhost:27017","XXX","XXX"]' ZBX_NOTSUPPORTED: timeout occurred $ zabbix_get -s 127.0.0.1 -p 10050 -k 'mongodb.server.status["tcp://another.working.mongodb:27017","YYY","YYY"]' ZBX_NOTSUPPORTED: timeout occurred $ zabbix_get -s 127.0.0.1 -p 10050 -k 'agent.hostname' my-hostname
Note: this only concerns the metrics of MongoDB plugin, the others continue to work.
Also these metrics work fine if I try them at the same time on another zabbix agent that hasn't encountered the bug yet.
Problem, when I set "DebugLevel=5" I can't reproduce the bug anymore (maybe it's a concurrency bug and logging takes enough time to reduce the probability of it occurring?).
And I can reproduce again immediately when I reset to "DebugLevel=3".