pcp
[Top] [All Lists]

RE: [pcp] Recent QA failures

To: "'Nathan Scott'" <nathans@xxxxxxxxxx>
Subject: RE: [pcp] Recent QA failures
From: "Ken McDonell" <kenj@xxxxxxxxxxxxxxxx>
Date: Tue, 28 Apr 2015 13:18:57 +1000
Cc: "'pcp developers'" <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1699060067.7689027.1430179931519.JavaMail.zimbra@xxxxxxxxxx>
References: <026e01d07fa2$874e9cd0$95ebd670$@internode.on.net> <1699060067.7689027.1430179931519.JavaMail.zimbra@xxxxxxxxxx>
Thread-index: AQFQrypyEwzA5vM7WhRPWDWLAh5YWAGlGsM2nlQ6BDA=
> -----Original Message-----
> From: Nathan Scott [mailto:nathans@xxxxxxxxxx]
> Sent: Tuesday, 28 April 2015 10:12 AM
> ...
> 995 looks like either the "python-pcp" or "python3-pcp" package has not been
> updated?  Its passing here for both python2 & 3 and I'd expect the failure
> you're seeing there if the pmapi.py was out-of-date.

This one (like Martins' ones) seems to have mysteriously gone away.

> 458 looks like either a pmdalogger issue or a race condition - nothing I can 
> think
> of that has changed there recently, so odd that its started failing.

This one is very strange.

For the regular file case there are 2 records in the file, there are 2 records 
Appended, Added and Removed via the libpcp_pmda Queue services (as reported by 
-Dlibpmda for the PMDA).

But pmevent in the qa test sees 3 (!) records,  and the last one (after the 2 
expected records) is the empty string "" ... this smells more like an EOF 
problem, rather than a race at start up.

But, wait, there is more ... with diags enabled (-Dlibpmda) for the PMDA the 
test passes ... disable the diags and it fails (I've repeated both experiments 
several times).

Not sure where to look now .... suggestions would be most welcome.

<Prev in Thread] Current Thread [Next in Thread>