pcp
[Top] [All Lists]

Fwd: Metrics With More That One PMID

To: PCP Mailing List <pcp@xxxxxxxxxxx>
Subject: Fwd: Metrics With More That One PMID
From: Dave Brolley <brolley@xxxxxxxxxx>
Date: Wed, 24 Feb 2016 12:33:37 -0500
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <56CDE8D4.5000308@xxxxxxxxxx>
References: <56CDE8D4.5000308@xxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.3.0
Sent this using the wrong email address and it was rejected by the list ...

-------- Forwarded Message --------
Subject: Metrics With More That One PMID
Date: Wed, 24 Feb 2016 12:31:00 -0500
From: Dave Brolley <dave.brolley@xxxxxxxxxx>
To: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
CC: PCP Mailing List <pcp@xxxxxxxxxxx>


Hi Ken,

Waaaaaay, back when we first started talking about this, you advised me to check the multi-archive meta data for
  • metrics with more than one PMID
  • PMIDs with more than one metric name.
While looking into this, I realized that the latter case is explicitly supported in the meta data file format. That is, we support PMIDs with more than one name. Was there some other issue that you had in mind with regard to this?

The former case, is flagged as an error by AddPMNSNode(), but the error is currently tolerated by __pmLogLoadMeta() -- see line 250 of logmeta.c. The comment there is incorrect. PM_ERR_PMID is returned when the same name has a different PMID. Is there some reason why this would a problem in particular for a multi-archive context?

FWIW, there is a a test case (507) for which the metric pmcd.pmie.configfile has a different PMID in the given name space file (src/root_irix) than it does in the given archive (archives/pcpcmd). This test started failing when I added the check for metrics with more than one PMID.

Thanks for your advice,
Dave


<Prev in Thread] Current Thread [Next in Thread>
  • Fwd: Metrics With More That One PMID, Dave Brolley <=