Schedules not running after SIA restart in BI 4.2 SP2 Patch 1

[former_member369090]

In a clustered environment, during a failover, all the BO servers i.e Server 1, Server 2 and Server 3 were restarted. However, on one of the servers schedules did not get assigned and did not run. They were manually assigned and gets into Pending state.

Also, servers like Adaptive Job Server, Crystal report server, Dashboard Server, connection Server get into disabled state and had to manually enable them.

Environment - Production

BI Version - 4.2 SP2 Patch 1

It would also be helpful, if there is best practices in restarting servers on a clustered environment.

Thanks

[denis_konovalov]

there is not enough information in your question to provide any answer.
What failover process ?
What sort of schedules run on one server only ? what does that even mean ? (in a cluster schedules run on any server that has job/processing servers required for the job)

I suspect you have a Disaster recovery system with 2 environments and that the failover mechanism you use is not working well.

Generally to restart a cluster first you stop all servers, make sure no left over process are running, then start Nodes with CMS servers first, when CMS nodes are running and CMS servers start to accept requests (logons for example) you start all the rest of the nodes.

[former_member369090]

Hi Dennis! The production environment is a clustered setup where there are 4 nodes and each one has a CMS. There are no issues when the SIA are restarted in sequence say Server1, 2, 3 and then 4. However, if no sequence is followed while restarting SIA there are issues like schedules are not running or getting into pending state.

Also, some of the services mentioned above are disabled and do not come up automatically when SIA starts.

So wanted to know if there are any recommendations from SAP on the sequence of restarting SIA on a clustered environment or is this an expected behavior?

[denis_konovalov]

how many concurrent users do you have and what's the total number of objects in your CMS DB ?
4 CMS servers is quite an over-kill for most deployments and creates a lot of CMS to CMS chatter.
This explains why schedules do not start right when you start all at the same time as CMS's are trying to figure out too many things at the same time between 4 of them....
as to the disabled services that don't start right -- this could be resources thing, or related to CMS being too busy.
Try to start first node, wait 10 minutes, start second , wait 10..... etc...

p.s.

Why do you need to restart your cluster often ?

[former_member369090]

The Servers are restarted as part of the scheduled server activity. They are restarted using scripts in the sequence - each server is restarted at an interval of 15 minutes. ANd there are no issues in this scenario where the servers are restarted in a sequence at a time interval.

When there is an issue with the Windows server and the server goes down for some reason, the other CMS has to be manually restarted. If the manual restart is done in the sequence of server 1, 2, 3 and 4 there are issues of servers getting into disabled state, schedules and publications getting stuck in running state.