Problem in starting SAP liveCache server

[Former Member]

Hi,

I am getting the following error message when i tried to start SAP liveCache server with LC10.

"Restart of liveCache LC3 on saplc3cache terminated; reason: User LC3RFC has lock"

Message no. LVC072

Does anyone has idea! abt the error? why it comes and how it can be solved?

Thanks,

Venkat

[Former Member]

Hello Venkat,

could you give us some more information. We need to have knldiag to check if the liveCache could be started and the error was forced by LC10 because of ABAP Problems. Could you please send us knldiag and lcinit.log file.

Regards, Christiane

[Former Member]

Hi Christiane,

The startup procedure is not even reaching the lcinit script. There are absolutly no logs in knldiag and lcinit.log files. After little bit debugging I found the problem. But don't know how to solve it.

The problem occured when I was doing disrptive testing for the liveCache server.

When I kill ("KILL -9 PID") the liveCache server processes (<b>master first and then child</b>) there are some IPC resources hanging over there. These IPC resources (shared memory segments) holds some locks and are preventing the liveCache server to startup.

The problem is not happening when I kill child first and then master process (all the IPCs are getting removed in this case).

I am wondering whether SAP has any tool to remove these IPC resources for liveCache server uniquely for perticular liveCache server (something similar to cleanipc for SAP systems).

I can't write a script to remove all the IPC resources based on the user name ('sdb') which same for liveCache server and MaxDB. (Imagine a case where MaxDB(APO DB) and liveCache server are running together). Any other solutions?

Any answers/suggestions are most welcome.

Thanks,

Venkat

[Former Member]

Hello Venkat,

which liveCache version do you use?

Regards, Christiane

[Former Member]

Hi Christiane,

I am using SAP liveCache version 7.6 with build level 22.

Thanks,

Venkat

[Former Member]

Hello Venkat,

you wrote that your first test failed when killing ("KILL -9 PID") the liveCache server processes (master first) there are some IPC resources hanging over there. These IPC resources (shared memory segments) holds some locks and are preventing the liveCache server to startup.

You killed the COORDINATOR process. The COORDINATOR process is the first thread which is started when you start the database instance. The COORDINATOR coordinates the start of all other database threads. The Coordinator controls all Threads of the database instance. When the database got erros the COORDINATOR is responsible for stopping all other database threads. When the database crashes the IPC-Ressources will be cleaned by the Coordinator process.

To simulate a liveCache crash you should kill the child process then the coordinator will clean up the other ressources.

You will find more information about the liveCache Threads in the MaxDB Documentation. I'm creating a new FAQ note which answers questions about MaxDB/liveCache Runtime Environment. Hope this note will be avaialble in english soon.

Regards, Christiane

[Former Member]

Hi Christiane,

Thanks for your prompt reply. There is no problem when I kill the child first (master will die almost immediately and automatically). The problem is when for some reason if the master/COORDINATOR process gets killed (In this case child processes won't die and will be hold by init process as it's parent) then liveCache server still functions. Now if I kill the child and restart the liveCache is not coming up. I need some mechanism for cleaning the shared memory for in this scenario.

Do you have any idea about how to do clean the shared memory in the above scenario (not manually)? Or there is no way(documented) you can cleanup the shared memory?

This is very important for me in HA setup. The cluster should be able to restart the server in case of these scenario without administrator's intervention.

Any help or suggestion will greatly help me.

Once again thanks for your help.

Thanks,

Venkat