[ClusterLabs] crm resource trace (Was: Re: trace of resource - sometimes restart, sometimes not)

[Reid Wahl]

On Fri, Oct 7, 2022 at 6:02 AM Lentes, Bernd
<bernd.lentes at helmholtz-muenchen.de> wrote:
>
>
>
> ----- On 7 Oct, 2022, at 01:18, Reid Wahl nwahl at redhat.com wrote:
>
> > How did you set a trace just for monitor?
>
> crm resource trace dlm monitor.

crm resource trace <rsc> <op> adds "trace_ra=1" to the end of the
monitor operation:
https://github.com/ClusterLabs/crmsh/blob/8cf6a9d13af6496fdd384c18c54680ceb354b72d/crmsh/ui_resource.py#L638-L646

That's a schema violation and pcs doesn't even allow it. I installed
`crmsh` and tried to reproduce... `trace_ra=1` shows up in the
configuration for the monitor operation but it gets ignored. I don't
get *any* trace logs. That makes sense -- ocf-shellfuncs.in enables
tracing only if OCF_RESKEY_trace_ra is true. Pacemaker doesn't add
operation attribute to the OCF_RESKEY_* environment variables... at
least in the current upstream main.

Apparently (since you got logs) this works in some way, or worked at
some point in the past. Out of curiosity, what version are you on?

>
> > Wish I could help with that -- it's mostly a mystery to me too ;)
>
> :-))



-- 
Regards,

Reid Wahl (He/Him)
Senior Software Engineer, Red Hat
RHEL High Availability - Pacemaker