Date: Mon, 20 Oct 2014 11:31:13 +0300 From: Konstantin Belousov <kostikbel@gmail.com> To: Jim Pazarena <fports@paz.bz> Cc: current@freebsd.org, freebsd-ports@freebsd.org Subject: Re: at - atrun utility bug on 10.0 Message-ID: <20141020083113.GL2153@kib.kiev.ua> In-Reply-To: <294ded22fe13975898c8d5e4b1a79a01@paz.bz> References: <294ded22fe13975898c8d5e4b1a79a01@paz.bz>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Oct 19, 2014 at 10:53:35AM -0700, Jim Pazarena wrote: > I find that while a job is running via the at facility, that is, > atrun has executed it, every subsequent execution of atrun (via > cron) STAYS running while the original program (executed by atrun) > is still active. > > Generally, according to the docs, atrun is executed every 5 minutes > */5 * * * * within /etc/crontab > I always modify this to * * * * * for more prompt at execution. > > I have jobs submitted via 'at' which tend to run for several hours > sometimes even all day. > If I do a 'ps wwax' I could see dozens, hundreds, even thousands > of "atrun" running. > When the original job completes all the "atrun"s disappear. > I noticed this on 10.0, where on 9.1 it simply does not happen. > > It is my feeling that the atrun on 10.0 is either being held up by > a lock fyl, an flock, or something else which is blocking its > normal termination when there is an empty queue, or rather when > there is an item in the queue which is already being tended to > by another process. > > I would think that the general */5 granularity of atrun, along with > most jobs NOT running for multiple hours tends to obfuscate this > (what I think is a) bug. And most people looking at an wwax would > skip past it without giving it any thought. > > If anyone could confirm that I am not going insane, I would file a > bug report thru normal channels. Of course, I may be going insane > regardless of THIS email :-) I doubt that port people can help you. What version of FreeBSD do you run, exactly ? It is 10.0 or stable/10 ? It seems that your observations are sound, and your trouble is caused by HEAD r251625. OTOH, it seems that HEAD r264617, merged to stable/10 as r265368, fixed something which is described very similar to the bug you hit.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20141020083113.GL2153>