Forum Discussion

dansharp_256889's avatar
dansharp_256889
Icon for Nimbostratus rankNimbostratus
Nov 13, 2018

MCPd Heartbeat

All,

Currently running into a bug where we have too many objects and a work around provided was to disable the MCPD heartbeat between our LTMs. Can someone give me a deep dive into the MCPD process and the pros/cons of disabling the heartbeat that people have experienced from a real world perspective?

Please advise and thank you,

 Dan
  • Hey Dan

     

    I'm guessing it's the following bug you're hitting?

     

    Bug ID 642923: MCP misses its heartbeat (and is killed by sod) if there are a large number of file objects on the system

     

    Disabling the heartbeat timer means that, should the mcpd process legitimately become unresponsive, the system will not automatically restart mcpd to recover.

     

    Like the article states, if something were to happen with the MCPd process, the system would not detect it and automatically restart it. I have no experience of turning it off but MCPd is one of the major deamons in TMOS and you will have nasty issues appear if it's down. Here is a description of the daemon:

     

    Name: mcpd

     

    Description: The master control program daemon is the messenger service that allows two-way communication between userland processes and the Traffic Management Microkernel (TMM).

     

    Impact if not running: No traffic management functionality; the system status cannot be retrieved or updated, and the system cannot be re-configured; other daemons will not be functional /var/log/ltm

     

    Source: K89999342: BIG-IP daemons (12.x)

     

    However, since it is only related to the number of objects and the mcpd taking longer than 300 seconds to finish its task, I'd say it is not critical but I would add a bug fix as soon as I can because I would have a bad feeling in my gut to have the heartbeat turned off for such an important daeamon.

     

    I hope this helps!