-
Type:
Incident report
-
Resolution: Cannot Reproduce
-
Priority:
Major
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
Situation
We have a set-up running for which there's a VPN tunnel from all our clients towards a central zabbix server. This zabbix server is not accessible from the outside world; only via the tunnel.
Observation
It seems at some point a packet is lost, after which the proxy immediately states that there is a connection issue and it will try to recover from this. However, although the connection is stable and configuration data is downloaded, still the connection state machine thinks are not okay.
Only restarting the zabbix proxy actually recovers from that situation.
Steps to reproduce (below is a way for which I found the issue to be always reproducible it is not our production set-up whatsoever):
- Set-up a host running CentOS (only tried using CentOS so might not be relevant)
- Install OpenVPN server and disable re-use of client credentials
- Set-up another host running CentOS
- Install OpenVPN client and configure 2 configurations towards the same server (basically copy the client config...)
- Start both tunnels
- Install zabbix proxy and configure as active node
- Observe the web admin panel > administration > proxies and see how the "last seen" column for that proxy increases to 2 minutes
- Observe the log on the proxy side for connection issues
Since this seems like a packet drop situation, I can imagine that it could also be reproduced by physically disconnecting a cable and plugging it back in some short amount of time later.
*Result:* _*Proxy Log file:*_ 31576:20190527:064902.316 Still unable to connect... 31576:20190527:065110.700 Still unable to connect... 31574:20190527:065207.027 received configuration data from server at "<private-address>", datalen 119793 31576:20190527:065319.084 Still unable to connect... 31576:20190527:065527.340 Still unable to connect... 31576:20190527:065735.596 Still unable to connect... 31576:20190527:065943.852 Still unable to connect... 31576:20190527:070152.108 Still unable to connect... 31576:20190527:070400.364 Still unable to connect... 31576:20190527:070608.750 Still unable to connect...
Expected:
I expected that the connection would not be lost or it would correctly identify that the problem is resolved