[pcp] pmcd gets stuck with pmda kill

[Frank Ch. Eigler]

Martins Innus <minnus at buffalo.edu> writes:

> [...]
> #################
> delta = 1 min;
>
> some_inst (
>     pmcd.agent.status != 0
> ) -> shell 10 min "pmsignal -s HUP -a pmcd"
>   & syslog 10 min "Restart unresponsive PMDAs" " pmda%i[%v]";
> #################

FWIW I'm not a fan of this approach, for a couple of reasons.

- it requires a separate process to be running & polling

- the polling implies a relatively slow response time, and a
  low-rate-limited response

- it cannot operate remotely (since pmsignal doesn't work across
  the network), thus can't be default-on in pmieconf

- should it happen that there are other pmcds running, for testing
  or whatever reasons, pmsignal will signal them all; we already
  have similar problems with the testsuite's and the rc.d scripts' 
  pmsignal calls killing unintended processes

I wonder why this seems in any way preferable to teaching pmcd or
pmdaroot to auto-restart failing pmdas?  They're at the right
place at the right time.


- FChE