#2813 lck: lcknd healthcheck times out

[Alex Jones]

lcknd processes mbx events in a loop. Therefore, if local events are sent continuously this will become an infinite loop, and AMF health check will get starved out.

[Sérgio Lopes Marques]

Is there any workaround for this defect?

[Alex Jones]

Can you post traces from lcknd for this? I haven't been able to reproduce it for a while.

[Sérgio Lopes Marques]

Many thanks for your help.
You can find in attach the traces requested.
Please don’t hesitate to contact me in case you need something more.
Thanks again

[Alex Jones]

Try out this patch, and let me know how it does.

[Sérgio Lopes Marques]

Hi Alex,

Thanks a lot for your help.
I continue to reproduce the problem.
I am using a system with an opensaf cluster that has 2 controllers and 4 of 20 possible payloads.
Than with the help of a ruby script I am always rebooting randomly one of the controller boards, waiting for the just rebooted board to come up and then randomly rebooting again one of the controller boards and so on…
After some dozens of reboots the problem appears.

Thanks again and regards,

[Alex Jones]

Hi Sergio. I have been able to reproduce the problem. Try out this patch. It includes the previous one as well as some others.

[Sérgio Lopes Marques]

Hi Alex, It's working! Many thanks

[Alex Jones]

Hi Sergio. I did some more testing, and found a minor issue with the patch. I've tested it, and it still works in my setup. Please try this new one on your setup.

[Sérgio Lopes Marques]

Hi Alex. This last patch also works in my setup.
Many Thanks

[Sérgio Lopes Marques]

[Alex Jones]

commit 876fbce762044d49da8edbd6bfcb059ee59e748e (HEAD -> develop, origin/develop, ticket-2813)
Author: Alex Jones ajones@rbbn.com
Date: Mon Feb 3 08:49:32 2020 -0500