November 5, 2015 at 8:01 am
Simple question for you experts 🙂
We have a active\passive 2012 SQL install using 2012 FOCM.
If the SQLSERVER service fails\stops on the active node, should this initiate a failover to the passive node?
Thankis in advance
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
November 5, 2015 at 9:09 am
What's FOCM? I can't find that acronym
Gerald Britton, Pluralsight courses
November 5, 2015 at 9:13 am
g.britton (11/5/2015)
What's FOCM? I can't find that acronym
Sorry Gerald, Windows Server Failover Cluster Manager
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
November 5, 2015 at 2:16 pm
So you have Windows server failover clustering, running a clustered instance of SQL? The default config for SQL seems to be 1 failover in a 6 hour period. and for the SQL service attempt to restart on the current node if it fails.
So no, if you do a simulate failure on the SQL server services the cluster should just try restarting them on the current active node.
November 6, 2015 at 1:33 am
PretendDBA (11/5/2015)
So you have Windows server failover clustering, running a clustered instance of SQL? The default config for SQL seems to be 1 failover in a 6 hour period. and for the SQL service attempt to restart on the current node if it fails.So no, if you do a simulate failure on the SQL server services the cluster should just try restarting them on the current active node.
Thank you for your reply. Yes the above clustering is the case with at present one instance. So if the SQLSERVER server service fails to start on the active node will an attempt be made to failover to the passive node?
Attached is a screenshot of the policy of the resource
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
November 6, 2015 at 4:04 am
Have you done testing?
November 6, 2015 at 5:36 am
PretendDBA (11/6/2015)
Have you done testing?
Yes i have carried out testing and have encountered the following behaviour, can anyone tell me if this is expected behaviour.
If i stop the SQL Service from SQL config manager, the service stops and in failover cluster manager the resource stops and stays offline without the restart attempt or failover.
If i stop the service in "services" or end the process, failover cluster manager restarts the service. If i disable the service then stop the service, failover cluster manager fails over as it cannot be restarted on the active node.
My question now is why should stopping the service in sql config manager see different behaviour in failover cluster manager as opposed to stopping the service in services. Surely the healthcheck from cluster manager to the active node would still occur
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
November 6, 2015 at 7:28 am
mark.williams 37494 (11/6/2015)
PretendDBA (11/6/2015)
Have you done testing?Yes i have carried out testing and have encountered the following behaviour, can anyone tell me if this is expected behaviour.
If i stop the SQL Service from SQL config manager, the service stops and in failover cluster manager the resource stops and stays offline without the restart attempt or failover.
If i stop the service in "services" or end the process, failover cluster manager restarts the service. If i disable the service then stop the service, failover cluster manager fails over as it cannot be restarted on the active node.
My question now is why should stopping the service in sql config manager see different behaviour in failover cluster manager as opposed to stopping the service in services. Surely the healthcheck from cluster manager to the active node would still occur
A clustered SQL Server instance should only be managed via SQL Server configuration manager console or the Failover Cluster Manager console.
Stopping the service via services.msc will restart the instance on the same node as will terminating the EXE process, this is a software failure and as such a same node restart will be attempted. However, multiple failures that fall within the policy constraints will eventually restart on a partner node. Hardware failures will initiate a restart on an available partner node.
The default policy is somewhat restrictive, especially if you're carrying out tests, as reaching the max number of failures in the time frame allowed will affect the resources ability to come online, it's best to look closely at this policy and change it to meet your needs.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
November 6, 2015 at 7:35 am
Thank you for your reply Perry. I do understand what you are saying but why does'nt Failover Manager behave in the same way in terms of restart if possible or failover when the SQL Service is stopped using SQL config manager?
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
November 6, 2015 at 7:54 am
mark.williams 37494 (11/6/2015)
Thank you for your reply Perry. I do understand what you are saying but why does'nt Failover Manager behave in the same way in terms of restart if possible or failover when the SQL Service is stopped using SQL config manager?
When stopping via FCM or SQL Config manager it's deemed as because an administrator is forcing the resource offline, the resource will not restart until the administrator restarts it. Anything outside of these is deemed as an unauthorised stop and will be restarted.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
November 6, 2015 at 7:59 am
Perry Whittle (11/6/2015)
mark.williams 37494 (11/6/2015)
Thank you for your reply Perry. I do understand what you are saying but why does'nt Failover Manager behave in the same way in terms of restart if possible or failover when the SQL Service is stopped using SQL config manager?When stopping via FCM or SQL Config manager it's deemed as because an administrator is forcing the resource offline, the resource will not restart until the administrator restarts it. Anything outside of these is deemed as an unauthorised stop and will be restarted.
Thank you Perry, that fully explains it
_________________________________________________________________
"The problem with internet quotes is that you cant always depend on their accuracy" -Abraham Lincoln, 1864
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply