Automated reboot of OneClick and SpectroServer on Linux 7.6 server

[Aubrey Conner]

We have just migrated to Spectrum 10.4 onto a Redhat Linux 7.6 platform and my Digital Delivery guys are using Titanium agents and scripts to automatically force system and security patches monthly to the boxes.  Sometimes there are multiple patches with reboots inbetween.  I got lucky this weekend with finding out when it was going to happen and was able to bring it down gracefully before the patch but I may not be so lucky the next time and I don't want to babysit the process every patch cycle.  What is the best option to automate the shutdown and startup process?  I'm a complete novice in this process, so keep it simple.

------------------------------
Aubrey Conner
Systems Administrator
IT-Monitoring and Systems Management
RAYMOND JAMES & ASSOCIATES INC.
------------------------------

[Rajashekar Allala]

Hi Conner,

We have been using a script (to stop Spectrum) at OS run levels in RHEL 6.x. Not sure, if the same works in RHEL 7.x.




------------------------------
Thank you.
Rajashekar
------------------------------

[Joseph Ackley]

Aubrey,

The issue with shutting down the system is before processd can shutdown the SpectroSERVER and Archive Manager processes gracefully, the shutdown kills them.
On Windows, you can increase the WaitToKillServiceTimeout value using the Registry Editor. 

On Linux, add the "PROCESSD_SHUTDOWN_TIMEOUT=60000" parameter to the /opt/SPECTRUM/spectrum80.env file. A restart of processd needs to be done for this to take affect.

How high you increase them depends on how long it normally takes the SpectroSERVER and Archive Manager processes to gracefully shutdown in your env.

To automatically start up, edit the SS.idb and ARCHMGR.idb files and set AUTOBOOTSTART to "Y". A restart of processd is required for this to take affect.

It is NOT recommended to set the AUTORESTART to "Y" for the SS and ARCH.

Joe

[Rajashekar Allala]

Hi Joe,

Will it also take care of stopping the Spectrum along with the processd to avoid SSDB crash?

On Linux, add the "PROCESSD_SHUTDOWN_TIMEOUT=60000" parameter to the /opt/SPECTRUM/spectrum80.env file. A restart of processd needs to be done for this to take affect.

------------------------------
Thank you.
Rajashekar
------------------------------

[Joseph Ackley]

Yes.

[Joern Henrichs]

Hi Joe!

"PROCESSD_SHUTDOWN_TIMEOUT=60000" - is this 60 seconds?

Cheers!
Joern


------------------------------
Senior System Engineer
Computacenter
------------------------------

[Joseph Ackley]

Yes that is 60 seconds.

[Joern Henrichs]

Thanks Joe!

------------------------------
Senior System Engineer
Computacenter
------------------------------

[James Brunner]

Just to let you know, we have our servers up in AWS and our internal DevOps team don't always pass on the message when AWS wants to shunt the instances across to new tin. So, we set our timeout to 120000 (2 minutes) and this has worked fine for us for a number of years.

JB.