Hi All,
What *is* this gam_server? top shows that it uses 92-97% of my cpu. Is this needed?
Thanks,
Phil
On Don, 2004-11-04 at 08:19 -0500, Alan Cox wrote:
It shouldn't be eating CPU, can you get an strace of it doing that and file a bug
There are already two reports for this problem: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=132354 https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=137439
Tom
PS: Do you know how to get a coredump when a program has been started with ulimit -c 0? I guess this should be possible with gdb.
On Thu, Nov 04, 2004 at 03:25:27PM +0100, Thomas Zehetbauer wrote:
On Don, 2004-11-04 at 08:19 -0500, Alan Cox wrote:
It shouldn't be eating CPU, can you get an strace of it doing that and file a bug
There are already two reports for this problem: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=132354 https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=137439
strace is one way, another simpler way is indicated there: http://www.gnome.org/~veillard/gamin/debug.html#Debugging1 beware it can fill a disk really fast.
for some reason the tree representation a loop for some users after some random time. Why, how, I still didn't found... The code should not be racy but it seems something breaks that expectation :-(
Daniel
On Thu, Nov 04, 2004 at 01:07:14PM -0500, Daniel Veillard wrote:
for some reason the tree representation a loop for some users after some random time. Why, how, I still didn't found... The code should not be racy but it seems something breaks that expectation :-(
mount --bind will certainly create infinitely deep directory trees, but surely you are doing loop detection for this, for NFS and the like ?
On Thu, Nov 04, 2004 at 07:18:04PM -0500, Alan Cox wrote:
On Thu, Nov 04, 2004 at 01:07:14PM -0500, Daniel Veillard wrote:
for some reason the tree representation a loop for some users after some random time. Why, how, I still didn't found... The code should not be racy but it seems something breaks that expectation :-(
mount --bind will certainly create infinitely deep directory trees, but surely you are doing loop detection for this, for NFS and the like ?
Well we don't recurse in the tree, we have to check one level down each monitored directory but we don't implement recursive checking. I doubt it's cause of the problem people are seeing (normal desktop usage), but I write this down in the TODO to not forget those corner case, thanks !
Daniel
Hi,
gam_server is part of gamin and provides monitoring for changes in the filesystem.
Unfortunately there seems to be a bug leading to a loop being created in gam_server's internal tree causing it to use as much CPU as it can get.
From rom my earlier correspondence with Daniel Veillard I think can say that we will not get useful information from strace, what we need to trace this down further is a coredump.
Tom
-
Alan Cox -
Daniel Veillard -
Phil Savoie -
Thomas Zehetbauer