On 2014-03-06 14:00, Jan Larres wrote:
> Like I said I had to restart tmux so I can't test that at the moment,
> but I could try to reproduce it.
So when I do
set -g status-right "#(sleep 30)"
set -g status-interval 1
I see that tmux starts hoarding the sockets. So it possible that your
tmux-status is the culprit as you suspect although I haven't found any
obvious part in it which could hang in case of high load. If that is the
problem then maybe the hanged instances are still running if you haven't
rebooted your machine. Try "pgrep tmux-status" to check this. I'd
definitely suggest adding start/exit logging to your status script just
make sure we know it isn't that what hangs.
In any case, I think this is a bug which needs to be fixed.
But there's something peculiar in your case which isn't happening in my
repro:
On Wed, Feb 26, 2014 at 11:49:50AM +1300, Jan Larres wrote:
> tmux 6850 jan 340u unix 0xffff88004e367480 0t0 2886303145 socket
> tmux 6850 jan 341u unix 0xffff8800c5c577c0 0t0 2886303152 socket
> tmux 6850 jan 342u unix 0xffff88006b6c0080 0t0 2886303153 socket
> tmux 6850 jan 343u unix 0xffff88006b6c0400 0t0 2886303154 socket
> tmux 6850 jan 344u unix 0xffff88004e2e38c0 0t0 2886303155 socket
> tmux 6850 jan 345u unix 0xffff880012806880 0t0 2886303170 socket
> tmux 6850 jan 346u unix 0xffff88005406e740 0t0 2886303171 socket
> tmux 6850 jan 347u unix 0xffff88006cdf5bc0 0t0 2258021114 socket
> tmux 6850 jan 348u unix 0xffff88004e367800 0t0 2886303224 socket
> tmux 6850 jan 349u unix 0xffff88004e367b80 0t0 2886303225 socket
> tmux 6850 jan 350u unix 0xffff88005406eac0 0t0 2886303226 socket
> tmux 6850 jan 351u unix 0xffff8800c5c57b40 0t0 2886303227 socket
> tmux 6850 jan 352u unix 0xffff88006cdf54c0 0t0 2258021115 socket
> tmux 6850 jan 353u unix 0xffff88006cdf5840 0t0 2258021128 socket
> tmux 6850 jan 354u unix 0xffff8800accda040 0t0 2258021129 socket
The socket inodes are consecutive which isn't in my case thus this might
mean it's a different leak. If the tmux-status investigation fails I
suggest creating a debug tmux and running the server under "valgrind
--track-fds=yes" to determine where were the fds actually allocated in
order to narrow down the search area.
--
Balazs
------------------------------------------------------------------------------
Subversion Kills Productivity. Get off Subversion & Make the Move to Perforce.
With Perforce, you get hassle-free workflows. Merge that actually works.
Faster operations. Version large binaries. Built-in WAN optimization and the
freedom to use Git, Perforce or both. Make the move to Perforce.
http://pubads.g.doubleclick.net/gampad/clk?id=122218951&iu=/4140/ostg.clktrk
_______________________________________________
tmux-users mailing list
tmux-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/tmux-users
