You are not logged in.
Ex ante: The CPU is an Intel i5 650 @3.20GHz, all packages are up to date, different google queries came up with nothing recent, relevant or even Linux-related.
I recently came across a performance problem with tmux. It occurs in a certain scenario:
X
- Openbox or Fluxbox, no composite active
-- terminator or xterm
--- tmux
---- ssh session to remote server
----- something output heavy like rsync'ing lots of small filesAfter about 20 seconds, tmux and X both show 100% CPU in top/htop (one per core, I guess), both X and the tmux session become very unresponsive, to the point where I had to go to a TTY and kill X. The longer it runs, the worse it becomes, yet again to the point where I even cannot go to a TTY, bacause the keyboard is blocked. It does not happen, if ssh is not involved, like running an output-heavy rsync in tmux locally. The effect can be observed no matter if tmux is actively drawn (minimized, on a different workspace). The load does not go down, even if I manage to go to a TTY. If I kill X and attach to the same tmux session descibed above without X, tmux produces 30-60% load on one core.
I elegantly dodged the problem by starting screen sessions on the servers, running my rsync or whatever and detaching them, until they are done.
Does anyone know this issue? Any insights on analytics I can do, to find out whether this is a bug to report, or just a SOL and JWF moment?
EDIT: For the protocol, this is marked as solved, it turned out to be a defective CPU.
Last edited by Awebb (2012-10-19 10:10:15)
Offline
I read somewhere on this forum that using the compression option with ssh will only slow down (and put stress upon) your computer. IIRC this is especially true if you're on a local LAN. Please don't take my word as gospel because my memory has failed me more than once. ![]()
I mention this only because it would be an easy thing to try and may help later on, if not now.
Offline
Well, FWIW:
monsterwm -> xterm -> tmux -> ssh to my local server -> rsync -av (verbose) maildir folder to another test folder on the server -> no problem. My computer is many year old Pentium D dual core...much less than an i5! Have you tried with urxvt? Same test with urxvt worked fine too.
Scott
Offline
Thank you for your input, both of you.
I just tested this at home, different hardware but the same software scenario (maybe besides graphics kernel modules). It doesn't happen there, so it is limited to that particular machine. I will try urxvt tomorrow.
EDIT: See top post. Defective CPU.
Last edited by Awebb (2014-07-21 12:04:54)
Offline