I would have to agree. It sounds like a feature is broken and if it can be fixed in the next release without breaking anything else then it should be.

Andy

Robert Story (Coders) wrote:
Problem: The function used to fork and exec for the extensible tables waits for
the child to exit before reading the pipe for the output. Thus, if the child
fills the pipe and blocks, the agent ends up deadlocked.

Solution: read from the pipe while waiting for the child to exit. The was the
behavior in the 5.0.x and earlier lines.

Patch:
http://sourceforge.net/tracker/index.php?func=detail&aid=1052460&group_id=12694&atid=456380


Status: I'm in favor of it going in for 5.2, Dave wants to wait til 5.2.1. Summary of our exchange below:

---------------------------------------------------
Dave Wrote:

Dave> If it's due to the change in the handling of forking off sub-processes,
Dave> then it will have been broken since 5.1 - i.e. for almost a year.
Dave> If it's taken that long to come to light, then I don't really think
Dave> this counts as a "show stopper".

Robert> I consider any segfault or non-recoverable hang of the agent
Robert> a show-stopper.

I'd certainly agree that this is a significant bug, which definitely
needs fixing.  I wouldn't want you to get the wrong idea about that.

I'm just saying that (IMO) the seriousness of a given problem has to
be judged alongside the likelihood of it actually biting.  And the fact
that it's taken so long to come to light, would tend to suggest that
this isn't necessarily an "urgent" fix.

Dave> I'm very reluctant to jeopardize the 5.2 release at this late stage.
Robert> Ok. Well, a proposed 'official patch' for 5.2 can be found at
                [snip]

Well, the first thing to say is well done in tracking down this problem
and coming up with a patch so quickly.  If you're happy that this is
likely to fix the problem, then it's definitely worth applying it to
the 5.1.x line.  I don't see any reason to hold off there.

But having looked at the code of the patch, IMO it's suffficiently
non-trivial that it would be better applied to the 5.2 line *after*
the release freeze has been lifted.
   But of course I'm happy to defer to the general consensus, if that
prefers to apply this to 5.2 straight away.




-------------------------------------------------------
This SF.Net email is sponsored by:
Sybase ASE Linux Express Edition - download now for FREE
LinuxWorld Reader's Choice Award Winner for best database on Linux.
http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
_______________________________________________
Net-snmp-coders mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders

Reply via email to