Re: SQL 2000 Cluster recovery after ethernet switch outage

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

So it seems that if the switch is out for less than the restart periods of
the IP addresses, the next restart will work and we'll be running. On the
other hand using a long timeout will delay a cluster changeover if the
network interface on one node (rather than the switch) goes down.

Many thanks for the info.

---
Graham

"Geoff N. Hiten" <SQLCraftsman@xxxxxxxxx> wrote in message
news:uyCm$35ZGHA.1200@xxxxxxxxxxxxxxxxxxxxxxx
You will need to start the clustered resources manually.

What happens is when the switch is off., the NIC shows as unplugged to the
OS, which takes the associated IP addresses offline causing a resource
failure. If you do not power up the switch within a certain time period,
the retry count for the virtual server is exhausted and the cluster "gives
up" on restarting the SQL Server. When the resource is restored, you have
to restart the cluster group and the SQL server group. You should be able
to connect the cluster admin tool to the local node name (not the cluster
name, it is still offline) and restart the cluster service and the SQL
Service without rebooting the nodes.

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



"Graham Morris" <Graywing@xxxxxxxxxxxxxxxx> wrote in message
news:eON%23Qi4ZGHA.4160@xxxxxxxxxxxxxxxxxxxxxxx
I have a two-node SQL Server 2000 cluster. Each node has two network
interfaces - one interface on each is connected to its opposite number by
a crossover cable, and the other interface is connected to an gigabit
ethernet switch, to talk to the outside world. All standard stuff. (Both
nodes are connected to the same switch).

The problem we have is this: if the gigabyte switch is powered down, the
cluster does not recover when it is powered up again. The crossover
cable is never disconnected and the power to each node is not
interrupted. The only way to get the cluster up and running again is to
reboot the nodes.

I'm hoping that I can get the cluster to recover by correct configuration
via cluster administrator - but how? Any help greatly appreciated.





.

Relevant Pages

  • Re: 2-node cluster wont stop sqlservr.exe during user-initiated f
    ... It is the only supported way to change service accounts for SQL ... Microsoft SQL Server MVP ... > I'm concerned that the roaming profiles required for cluster access to ... Go ahead and switch accounts, save, switch back, save again and ...
    (microsoft.public.sqlserver.clustering)
  • Re: Questions about configuration - urgent
    ... I have a question about settings on a clustered instance of SQL Server ... In "Cluster Administrator", it is clear that our groups are not set up ... and the services do not restart on their ...
    (microsoft.public.sqlserver.clustering)
  • Questions about configuration - urgent
    ... I have a question about settings on a clustered instance of SQL Server ... In "Cluster Administrator", it is clear that our groups are not set up ... and the services do not restart on their ...
    (microsoft.public.sqlserver.clustering)
  • Re: Switch redundancy on heatbeat network
    ... on the ability of the cluster to use one or more of the 'public' networks ... I support the Professional Association for SQL Server ... supported on the heartbeat network one cannot simply point one NIC to switch ... excluded from the heartbeat network for good reason). ...
    (microsoft.public.sqlserver.clustering)
  • Re: SQL 2000 Cluster recovery after ethernet switch outage
    ... What happens is when the switch is off., the NIC shows as unplugged to the ... which takes the associated IP addresses offline causing a resource ... to restart the cluster group and the SQL server group. ...
    (microsoft.public.sqlserver.clustering)