Re: Problem Sun Cluster 3.2 with vote
- From: James <nomail@xxxxxxx>
- Date: Mon, 27 Aug 2007 08:52:29 +0200
Il Fri, 24 Aug 2007 15:21:18 -0400, Mike Delaney ha scritto:
On Fri, 24 Aug 2007 08:54:36 +0200, James said something similar to:
Il Thu, 23 Aug 2007 20:58:49 -0400, Mike Delaney ha scritto:
OK, so it looks like you've got Fibre attached storage, multiple
paths to each disk from each host, and you've got MPxIO enabled.
So far, so good.
Physically, what kind of storage is this, and what does the connection
path look like?
Fujitsu-Siemens FibreCat Sx80
Do the hosts connect to this array directly, or are there switches involved?
Do both nodes access the LUNs provided by the RAID controller over the
same arbitrated loop, or does each host have its own loop to access the
LUNs that the RAID controller has been configured to let it see?
The nodes are connect with dual connection through a switches to two
controller of the array storage.
In the past, I've had problems with SunCluster 3.1 where one node
goes down, and when it starts probing for disks at startup the other
node thinks its SCSI reservation has been stolen and panics. This
is apparently only an issue where multiple nodes in the cluster
talk to the disks over the same FC loop. The StorageTek 3510s in
JBOD mode are (or at least were) specifically not supported with
SunCluster due to this issue - add RAID controllers into the path and
everything's happy.
Your symptoms sound very similar.
When NodeA crashes, does NodeB panic immediately, or does it stay up
until NodeA reboots? What messages does NodeB log when it panics?
The NodeB panic immediately. Now i haven't the log because i have
reinstalled the machine.
.
- References:
- Problem Sun Cluster 3.2 with vote
- From: James
- Re: Problem Sun Cluster 3.2 with vote
- From: kolayshkin
- Re: Problem Sun Cluster 3.2 with vote
- From: James
- Re: Problem Sun Cluster 3.2 with vote
- From: Mike Delaney
- Re: Problem Sun Cluster 3.2 with vote
- From: James
- Re: Problem Sun Cluster 3.2 with vote
- From: Mike Delaney
- Problem Sun Cluster 3.2 with vote
- Prev by Date: Solaris 10 under VirtualPC - Reset network after restarting saved session?
- Next by Date: Re: Upgrade from E10K to SF25K
- Previous by thread: Re: Problem Sun Cluster 3.2 with vote
- Next by thread: Re: Problem Sun Cluster 3.2 with vote
- Index(es):
Relevant Pages
|