I'm looking at an exploitation guide to plan an automatic stop and start of DCRUM solution.
I have found a old topic about DCRUM startup order :
Windows components services : startup order
The startup order was :
What about the stop order ? 3 => 2 => 1 ? or no special dependencies ?
Is there an special order in case of a CAS Cluster ?
I think we should first start the primary CAS and then the secondary CAS, right ? or there is not special order ?
What about the probes ?
Solved! Go to Solution.
The startup order is important because certain components have dependency on others to start up.
The list above is still accurate. As none of the windows services will start without SQL being available. The rest then need the CSS to start up.
The exact order of step 3 is not important as hey have no direct dependencies on each other.
For stopping the services the order is not as critical. A good rule of thumb would be 3>2>1 but as i say it is not that important as if you stop SQL it will stop the others. What you should be careful of though is to not stop these services if the CAS is in the middle of any maintenance tasks such as the nightly tasks.
AMDs have no dependencies on any of the windows services so you can stop/start it in whatever order you like.
In a cluster it would be best practice to start the primary node first.
What i would ask though is why you want to so an automatic start/stop of the solution? This is designed to be an always on solution and should only really be stopped for maintenance or in case of issues.
Thanks for you answer.
I want to restart automatiquely because :
- the servers are running Windows....
- On probes, I want to backup the important files (see Admin Guide chapter 4 : Backup and Recovery Procedures), it reset the statistics (DH,...), load news keys, ...
Anyway, it is a usefull information for maintenance or exploitation tasks.