Marking ringid 0 interface FAULTY

[itvietnam]

Hi,

Our corosync is usually said this and recover after 1 second. may i know how to debug?

Test omping in 5 minutes and ok. No packet lost.



This is the test on ringid 1, they are also has the same problem with ringid 0.



Error of ringid 1 (10.20.30.0/24)

 

[itvietnam]

This is the test i performed on ringid 0 with history command:

 

[Stoiko Ivanov]

does omping with a high packet rate work?
omping -c 10000 -i 0.001 -F -q NODE1-IP NODE2-IP ... (from https://pve.proxmox.com/pve-docs/chapter-pvecm.html)

else - I would start recording the traffic on the ring with tcpdump and take a look at the dump with wireshark.

 

[itvietnam]

can you provide command then i will test on server

Thanks,

 

[Stoiko Ivanov]

The `omping` command invocations are explained in our documentation.

For the tcpdump, I'd do something like
`tcpdump -s0 -w corosyncproblem.pcap -ni $IFACE`, where `$IFACE` is replaced by the interface your corosync network is bound on.

after gathering enough data (the log saying that the ring became faulty and ok again) - stop tcpdump with Ctrl-C, take the resulting corosyncproblem.pcap and open it in wireshark.

 

[VitoHoang]

Here is our result of omping with a high packet rate. No packets loss and the ring not came to faulty during the test. Have done this test many time.

We attach the tcpdump result.
https://drive.google.com/file/d/1S7gqhpTJ-ysUT42rqU-9Ynf-JQegi22O/view?usp=sharing

 

[Stoiko Ivanov]

You could try to set corosync to log debug messages - and see if the logs show anything with more information