Stop clustered services before reboot?

[benjaminway]

I am looking to find if it is a good practice to stop clustered services before the servers in an environment reboot?  Given that clustered files for services, pgo files and QVWs are all located on another server than those using them is there any concern of corruption by not stopping services before the servers go through their weekly reboot?

Here is an example environment:

QVFILE1 - Server acting as a QV file repository hosting several windows managed shares that contain pgo files, QVW files, and clustered services for DistributionService and QlikViewServer

QVSserver1 - Three clustered QVS servers that use windows manged share QlikViewServer on QVFILE1

QVSserver2

QVSserver3

QDSserver1 - Two clustered QDS servers that use windows managed share DistributionService on QVFILE1.  QDSserver1 also runs QMS

QDSserver2

Even if these servers reboot at the exact same time on a weekly basis is it better to stop QlikView services on the whole environment before they go through their reboot process?

Thank you

[Anonymous]

Why the need for a weekly reboot ?

Notwithstanding that, I always close things tidily before rebooting anything.  I have seen pgo and qvd corruptions after a server crash, so a forced reboot could do the same.

[gregortvw]

(In a perfect scenario) all the services are terminated while shutting down the server and starting after OS start.

the mentioned coruptions appear if QVS or other service is currently writing the file (PGO or reloading the QVW)

I also terminate all the services via short batch before rebooting and after the startup there is another one, that checks whether everything is up'n'running and starts it if not.

[benjaminway]

Thanks for the info.  I ended up creating a PowerShell script to stop all the services for the environment a few minutes before their weekly reboot tasks kick off.  So this should help to prevent any corruption or files being in use when the servers begin their reboots.  We also have a monitoring tool that can check on specific services on servers and will restart them if they are stopped, so that helps to cover us after the reboots if something should go wrong on their startup.