Re: pmlogrewrite quirk/bug?

Hi Ken,

----- Original Message -----
> On 25/10/16 08:46, Ken McDonell wrote:
> > ...
> > I spoke too soon ... qa/827 is passing, but a bunch of other -g
> > pmlogrewrite tests are now failing ... I'll need to spend more time
> > understanding what's going on here before pushing any code changes.
> 
> OK, now I think I understand.
> 
> pmlogrewrite is actually correct and pmlogger is not quite correct and
> pmlogextract follows the pmlogger approach.  The problem is the
> "preamble" pmResult that contains pmcd.pmlogger.host, pmcd.pmlogger.port
> and pmcd.pmlogger.archive and more importantly these are defined over
> the instance domain 2.1.
> 
> For pmlogger, the preamble is written first, then all of the remaining
> pmResult records are written from the callback method ... and the
> temporal index is only updated in the callback method, which means the
> instance domain from the preamble does not trigger a temporal index write.
> 
> For pmlogrewrite the preamble record is treated the same as any other
> pmResult and the temporal index write logic applies.
> 
> The difference is is purely academic given that the only metadata
> between the offset 351 and 7893 in the meta data files are pmDesc
> records for the first real pmResult.  If you're scanning for an instance
> domain (the only time the metadata offsets in the temporal index is
> used) it makes no difference if you start at 132 or 7983, and indeed
> since the _last_ pmResult containing an instance domain is the preamble
> at 10:10:27.854 (the timestamp on the first temporal index entry) then
> the last temporal index entry will never be used when scanning for an
> instance domain.
> 
> So I think it is a "quirk" and I'll adjust the 827.out if you agree.

Sounds good.  Since these are checked-in archives (qa/archives/new_*) I
guess we could also use pmlogextract to remove the preamble metrics and
the tests would probably all dodge this benign behaviour.  Only reason I
suggest that is the test (827 anyway) is checking for and reporting "no
differences" ... so it may be a bit odd to the next person who writes a
pmdalinux pmlogrewrite rule there too I guess.

Either or, whichever way floats your boat.  :)

> I probably should write down how this archive stuff really works,

That'd be awesome.

> sometime before I die.

And that, much less so.

Thanks Ken!

cheers.

--
Nathan

-=-=-=-=-=-=-=-=-=-=-=-
pcp mailing list
[email protected]
https://groups.io/g/pcp/messages
-=-=-
Groups.io Links:

You receive all messages sent to this group.

View/Reply Online (#14606): https://groups.io/g/pcp/message/14606
View All Messages In Topic (5): https://groups.io/g/pcp/topic/2968875
Mute This Topic: https://groups.io/mt/2968875?uid=174580
New Topic: https://groups.io/g/pcp/post

Change Your Subscription: https://groups.io/g/pcp/editsub?uid=174580

Group Home: https://groups.io/g/pcp
Contact Group Owner: [email protected]

Terms of Service: https://groups.io/static/tos

Unsubscribe: https://groups.io/g/pcp/leave/354243/563757577/xyzzy
-=-=-=-=-=-=-=-=-=-=-=-