Heartbeat in CentOS 6.0

[13zai]

Hi All,

Recently I'm trying to setup a HA as the NTP server to my client, I've setup it and manage to work but there is few issues that bother me:

[1]
When HA001 have problem (let's say heartbeat's link down), it will fail over to HA002, everything are work perfectly.

But After I've connect back HA001 into network, and try to disconnect HA002 to check whether it will fail over to HA001, it is failed.

I've check through the log file and found this line:

Nov 21 15:06:24 HA001 heartbeat: [1779]: info: ha001 wants to go standby [foreign]
Nov 21 15:06:34 HA001 heartbeat: [1779]: WARN: No reply to standby request. Standby request cancelled.

Since like this is the reason that HA001 do not take over the resource after HA002 failed.

[2]
When I'm checking the log file, I found there is lot of error message as below:

Nov 21 16:13:25 HA001 heartbeat: [4072]: WARN: Message hist queue is filling up (376 messages in queue)
Nov 21 16:13:27 HA001 heartbeat: [4072]: WARN: Message hist queue is filling up (377 messages in queue)
Nov 21 16:13:29 HA001 heartbeat: [4072]: WARN: Message hist queue is filling up (378 messages in queue)

I've check through online, since like there is packet lost between HA001 and HA002, but I'm sure there is no connectivity issues as this 2 HA are direct connect to a dummies switches.

Anybody encounter this issues before? I'm out of idea how to solve it, hope you all give some idea so I can test with it. I've attach both log file in the thread as well. thanks in advanced.

Cheers,

Kc13

[13zai]

Hi All,

I managed to solved my issues [2] by turning off my iptables.

I've check through my iptables entry and ensure I'm permit UDP 694 can go through the iptables, but seen like not working until I'm totally turn off iptables entirely.

Cheers,

Kc13

[13zai]

Hi All,

I run case by case testing and found some interesting result as below:

[IMG][/IMG]HA error.jpg

When we shut down HA001 or created unexpected power off (unplug power cable of HA001), the HA work perfectly but when we only de-attach the cable or turn of interface, HA cannot work exactly the same. Anybody encounter similar issues?

Cheers,

13