Hi Michael
> - did you try another port in the second ring, not 5405?
Yes. 5406.
> - Does tcpdump show the packets in the line?
Yes.
On hostB I can see packets from hostA who has a "dead" corosync.
IP 123.201.179.21.5405 > 226.94.1.2.5406: UDP, length 376
Let's explain everything correctly because my previous post wasn't clear
and neat.
*Issue*
# corosync-cfgtool -s
Printing ring status.
Local node ID 399026368
Could not get the ring status, the error is: 6
*Context*
- My configuration works until I decide to use a second ring.
- To add a second ring, I add in corosync
+ a new interface (ringnumber: 1, different multicast IP + port)
+ a new parameter: "rrp_mode: active"
+ restart corosync, even reboot the instance
*Info*
- During boot time, I have plenty of lines like the following :
"[TOTEM ] Process pause detected for xxx ms flushing membership
messages."
+ with one working ring, it stops after ~ 500ms followed by pcmk info
+ with 2 rings, it stops at 25000ms and not pcmk messages follow
- Corosync doesn't stop/restart with 2 rings. It blocks on :
"Waiting for services to unload:... ".
it's hard to strace due to the multicast flooding
*Config*
- Corosync 1.1.2
- Pacemaker 1.0.6
- Centos 5.4
*Questions*
- What means error 6 ?
- What corosync is waiting for during boot time or when shutting down ?
Any ideas ? :)
-Thomas
_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais
