On 05/27/2011 03:48 PM, David Smith wrote:
> On 05/25/2011 06:13 PM, Nathan Scott wrote:
>> There are also a couple of outstanding changes - Arthur's pmlogextract
>> fix still to go in, and I'm still trying to find time to get a couple
>> more tweaks to pmdalogger... once those are in, I think we're done for
>> this round. Let me know if I've overlooked anything.
>
> These probably aren't critical, but I've got a couple of small tweaks to
> the pmdalogger.
>
> (I've been working on a bigger improvement to break returned events at
> newlines, but this appears to bring up a bug in the event handling logic
> where it loses the 2nd event the 1st time the logger is restarted.)
I finally figured this one out - it was a bug in libpcp_pmda that can
happen anytime an event record is greater than 512 bytes. Can someone
pull this?
Change committed to git://oss.sgi.com/dsmith/pcp.git dev
commit 53bbad5de7dbdb7febb9add8357c4d63f38a12c9
Author: David Smith <dsmith@xxxxxxxxxx>
Date: Thu Jun 9 14:03:34 2011 -0500
Be sure that check_buf() updates the event record pointer after a
realloc.
Without this change, if realloc() moved the buffer around (instead of
just lengthening it), the current event record pointer will point to
the old address, not the new. This can happen if an event record is
added, then two event parameter are added (the 2nd causing a
realloc()).
--
David Smith
dsmith@xxxxxxxxxx
Red Hat
http://www.redhat.com
256.217.0141 (direct)
256.837.0057 (fax)
|