pcp
[Top] [All Lists]

[Bug 1133] event.flags / event.missed "anonymous" metrics registered too

To: pcp@xxxxxxxxxxx
Subject: [Bug 1133] event.flags / event.missed "anonymous" metrics registered too late
From: bugzilla-daemon@xxxxxxxxxxx
Date: Mon, 04 Jan 2016 13:36:05 +0000
Auto-submitted: auto-generated
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <bug-1133-835@xxxxxxxxxxxxxxxx/bugzilla/>
References: <bug-1133-835@xxxxxxxxxxxxxxxx/bugzilla/>

Comment # 2 on bug 1133 from
> [...]
> Is a fetchgroup client likely to be asking for event.* metrics?

That would be the natural way of grabbing the flags / missed-event counter.  We
don't have sophisticated event consumers at all at the moment, but one can
imagine those being useful someday.


> If yes, is this likely to be explicit or are they expected to be found in a
> PMNS traversal from the root of the PMNS?

In the fetchgroup case, the former - so perhaps one could teach the
pmLookupName/Desc part about them, if we can't put them into the PMNS properly.
 The key is to have the names be accessible -before- the pmUnpackEventRecord*
call.

It would be possible for fetchgroups to paper over this problem by making a
dummy pmUnpack* call on a dummy pmResult* during the registration phase, just
to trigger the registration code.


> If you could give me some use cases, I'll see if there is (a) a different
> place to pre-emptively register the metrics that would work for you, but not
> burden everyone else [...]

FWIW, fetchgroups aren't that special - the way that these magic metrics were
added impedes normal pmapi flows too; forces name lookups to occur within a
decode loop.


(By the way, having you be the default assignee for new oss.sgi.com/bugzilla
pcp reports should probably be changed.  These problems are generally not in
code you wrote.)


You are receiving this mail because:
  • You are on the CC list for the bug.
<Prev in Thread] Current Thread [Next in Thread>