pcp
[Top] [All Lists]

Re: Next release update and pmdalogger merge

To: Nathan Scott <nathans@xxxxxxxxxx>
Subject: Re: Next release update and pmdalogger merge
From: "Frank Ch. Eigler" <fche@xxxxxxxxxx>
Date: Fri, 13 May 2011 08:12:44 -0400
Cc: David Smith <dsmith@xxxxxxxxxx>, Ken McDonell <kenj@xxxxxxxxxxxxxxxx>, Max Matveev <makc@xxxxxxxxx>, pcp <pcp@xxxxxxxxxxx>
In-reply-to: <1566158908.55479.1305286647155.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx>
References: <1586456728.55457.1305285330525.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx> <1566158908.55479.1305286647155.JavaMail.root@xxxxxxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.4.2.2i
Hi, Nathan -


> [...]  The one major thing I'm thinking of doing still, which
> warrants a bit of discussion, is adding a storable setting to switch
> off/on the event record metrics.  [...]

> So, what I'm planning is: - make logger.perfile.xxx.records "off" by
> default, and require a pmstore to enable them. [...]

How about making this default part of the logger pmda's configuration,
done on a per-file basis?  That way a user can expose non-sensitive
data without imposing special requirements on the consumer tools.

> Finally, I had some issues with the command-pipe mode of operation
> in pmdalogger ... have you seen those David/Frank?  Don't think I
> broke that with my changes, but could be wrong there - we seem to
> get EBADF always on a pipe fd read.  PCP QA test 457 shows off the
> issue, but a conf line like "pipe yes|" is enough to reproduce it.

Sounds like a bug.

> We'll need to switch from pipe2 to something portable in that code
> (that __pmProcessCreate we discussed earlier), but be good to have
> an understanding of this issue first if possible.

I didn't realize pipe2 was a linux invention.  Plain pipe() or
something else that wraps that shouldn't be a big deal; just a
fcntl(O_NONBLOCK) later.

- FChE

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