[Linux-HA] Heartbeat, ipfail and two ethernet cards
alanr at unix.sh
Wed Mar 2 07:28:35 MST 2005
Alexis Januskiewicz wrote:
> Hi everybody,
> I'm trying to test heartbeat with a squid cluster.
> I've understood than Heartbeat can't monitor resources (so, if squid
> crashes, nothing happened, except if I use "mon" or "ldirectord"), and
> also that I've to use IPFail to detect network failures.
> My network is :
> ______________ SW _____________
> | |
> | |
> |eth0 |eth0
> ___|___ ___|___
> | |eth1 eth1| |
> | SRV 1 |-----------------------| SRV 2 |
> |_______| Heartbeat link |_______|
> First time, I only used one ethernet card on each server. The problem is
> that if network link is broken, each server think he's alive and when
> link goes up, heartbeat restart on each server. Is it normal ? For me,
> it's not, but maybe I'm wrong.
> Consequently, I choose to dedicate a second network card to heartbeat,
> to avoid having lot of UDP traffic on my network and also to permit to
> monitor servers even if eth0 is down.
> First, without using IPFail, I unplug eth0 from SRV1. Heartbeat thinks
> that SRV1 is still alive, and doesn't switch VIP to SRV2.
> After that, I decided to "heartbeat" eth0 AND eth1, but I'm in front of
> the same problem.
> So, finally, I configured Heartbeat to use IPFail. The result is not
> better, and I really think that I've a problem with my configuration.
> I've two questions :
> - do you think this configuration is correct and "optimal" ? If not,
> what do you recommend ?
> - are IPFail logs normal ("respawning too fast") ??? If not, what could
> I do ?
No, these messages are definitely broken...
> heartbeat: 2005/03/02_14:35:04 info: Starting "/usr/lib/heartbeat/ipfail" as uid 1002 gid 104 (pid 16275)
> heartbeat: 2005/03/02_14:35:04 WARN: Exiting /usr/lib/heartbeat/ipfail process 16275 returned rc 126.
> heartbeat: 2005/03/02_14:35:04 ERROR: Client /usr/lib/heartbeat/ipfail "respawning too fast"
But it is not at all obvious what would cause it to exit with return code
of 126. I can't find this anywhere in the code...
Alan Robertson <alanr at unix.sh>
"Openness is the foundation and preservative of friendship... Let me claim
from you at all times your undisguised opinions." - William Wilberforce
More information about the Linux-HA