Re: Passive node starting the SQL Server services - services set t

You are very good my friend. Yes, this is an HP BL40p blade cluster. It
causes me great pain. It is an inherited environment. I take it blade
clusters have issues..??

"Geoff N. Hiten" wrote:

Yes. Node 1 is running the SQL Virtual Instance. Node 2 then becomes the
"watchdog". Node 2 suddenly gets no response from Node 1. Node 2 then
tries to start the SQL virtual server but Node 1 will not release the disk
resources. Node 2 then sees Node 1 again with the Virtual Instance up and
running so it stops trying to take over. Node 2 does not know why it cannot
talk to Node 1, only that the other node is non-responsive. My guess is
that you lost both the crossover connection and the public connection at the
same time. This isn't a blade server based cluster, is it?

--
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP




"Christopher" <Christopher@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:13131751-3C22-4097-A472-06C42027EFD2@xxxxxxxxxxxxxxxx
So, if the node is the passive node and has connectivity issues, would
that
cause this problem? Remember, the passive node was having issues.

"Geoff N. Hiten" wrote:

Failover and failback settings are for the instance, not the node, as you
discovered.

Possible owners list is not sorted, I.E. the order does not matter.

Preferred owners list IS sorted by most preferred owner first.

If you were having problems with internet connectivity, the system may
have
lost the IP ADDRESS resource long enough to start a failover or restart
process. Bad hardware can cause some wierd cluster instabilities.

--
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP


"Christopher" <Christopher@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:DD4224CF-4748-437F-AECA-11C23AF759B2@xxxxxxxxxxxxxxxx
Geoff,

I'm not sure what you mean in regards to failover/failback option for
each
node. Going to properties of each node yields only one tab. As
mentioned
below, the preferred owners for the resource group and possible owners
for
the resource were inverted:
Preferred owner(s)
NODE1
NODE2

Possible owner(s)
NODE2
NODE1
(someone changed the above. Now, both settings are the same).
Could you point me to the area you are referring to below - preferred
node
settings?

The resource group has the "prevent failback" radio button selected.

The application logs on both machines have the same information: the
data
drive not being available - the server name is the SQL Server Network
Name.

It turns out that there was a problem with backplane of this server
which
controls internet connectivity. After this hardware was replaced and
the
server rebooted, the problem ceased. I'm not sure if that has anything
to
do
with the resolution.

Thanks in advance.
-C


"Geoff N. Hiten" wrote:

Manual start for services is correct for a cluster. The cluster
should
control the services on each node. Look at the applicaiotn logs on
both
nodes. Check the failover/failback settings. You may be seeing a
failback
condition due to different preferred node settings.

--
Geoff N. Hiten
Senior Database Administrator
Microsoft SQL Server MVP




"Christopher" <Christopher@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:E5873D57-2A8F-4B7B-B76C-7146B8A629DD@xxxxxxxxxxxxxxxx
Hello all,

I have a very bizarre situation with a single instance/two node/W2K3
cluster. The issue I am seeing is that the passive node is starting
the
sql
server services even though the resources are not owned by that
node.
The
restarts are occuring every five minutes. The system log shows the
SYSTEM
account starting MSSQLSERVER. Then, 3 seconds later, I get a
message
stating
the service stopped. About 5-8 seconds later, I get a message that
the
SQLSERVERAGENT could not be started because sqlservr.exe is not
running.
The
cluster log is not showing anything being started/stopped. The
applicaiton
log is complaining about the errorlog being unavailable. However,
this
is
correct because the resources are not on this node.

The SQL Server services are set to manual on both nodes. Both nodes
have
been rebooted in the past 24 hours due to EMC PowerPath upgrades.
The
only
thing that is a little bizare is that the SQL Server resource group
and
SQL
Server resource have the preferred owners/possible owners in reverse
order.
IE (SQL Server resource group has NODE1/NODE2; SQL Server Resource
has
NODE2/NODE1). Right now all resources are on NODE1.

Any ideas would be greatly appreciated.

Thanks in advance
Chris.









.

Loading