[breezy] gamin has leaked 1.5gigs in the past 3 days

Bug #19692 reported by desrt on 2005-08-14
60
Affects Status Importance Assigned to Milestone
gamin (Ubuntu)
High
Sebastien Bacher

Bug Description

I have 2 gigs of ram and I noticed that for some reason 700megs of my swap space
was used. This confused me so I started looking for answers.

desrt 6333 0.0 44.6 1486744 927136 ? S Aug10 0:14
/usr/lib/gamin/gam_server

Sebastien Bacher (seb128) wrote :

Thanks for your bug. Are you using dnotify or inotify? What version of the package?

desrt (desrt) wrote :

Default config, so I'm guessing inotify (valgrind complains about unknown
syscall 291, which is an inotify syscall).

Version: 0.1.5-0ubuntu1.

I've been valgrinding it all night (since I filed the bug) but valgrind hasn't
produced any output. I'm going to see if Kjartan can give me some hints.

Sebastien Bacher (seb128) wrote :

thanks for working on that

desrt (desrt) wrote :
Sebastien Bacher (seb128) wrote :

*** Bug 20636 has been marked as a duplicate of this bug. ***

desrt (desrt) wrote :

Ross: if you can get a valgrind log of the leak it would be massively
appreciated (since nobody else is able to get any traction on this at all and
it's a significant issue).

desrt (desrt) wrote :

Just going through 'my bugs'...

I understand that upstream thinks that they've probably fixed this in CVS. I
don't know if they've done a release or anything, though....

Sebastien Bacher (seb128) wrote :

This bug is supposed to be fixed with 0.1.6 ... version which already has a
"Large revamp of the inotify back-end" according to the NEWS file, so something
we probably don't want to push just before a new stable. fcrozat backported the
leak fix to 0.1.5 for mandriva but rolled it back then since it seems to create
some crash with it ...

John Clemens (clemej) wrote :

Just a ping on this one.. I've had to kill gam_server twice in the past 4 days
for eating up all my memory.. While I agree this might be a little
destabilising, having something that you know is wrong and it going to lead to
systems getting obnoxiously slow is also not cool. Right now, it's something
I've started having to check daily. I'm just running a standard install.
Please consider pushing 0.1.6 to breezy.

Sebastien Bacher (seb128) wrote :

*** Bug 22840 has been marked as a duplicate of this bug. ***

Sebastien Bacher (seb128) wrote :

the patch fixing that for 0.1.6 has some crashers ...

John Clemens (clemej) wrote :

Hmm... good point, so how do I disable gamin? I use gnome as a glorified window
manager.. is there any way to work around this, revert it, or otherwise dissable
it until its fixed? what gnome program uses it?

Sebastien Bacher (seb128) wrote :

you can boot with "noinotify" to use dnotify which doesn't have this issue

John Clemens (clemej) wrote :

(In reply to comment #13)
> you can boot with "noinotify" to use dnotify which doesn't have this issue

thanks, that's all I needed.

Roland Dreier (roland.dreier) wrote :

What is the plan for handling this in breezy? I don't see how the release can
ship with a leak like this.

Matt Zimmerman (mdz) wrote :

(In reply to comment #15)
> What is the plan for handling this in breezy? I don't see how the release can
> ship with a leak like this.

It only seems to affect a few users, the cause is unknown, and there is a simple
workaround available. If someone can track down the leak, it's possible that we
can squeeze in a fix before the final release, but it won't be delayed on
account of this bug.

Roland Dreier (roland.dreier) wrote :

(In reply to comment 16)

Understood that you don't want to hold the release for this.

However it seems to happen on my system every time. Booting with "noinotify" is a
pretty big stick -- for example, that kills using beagle.

It seems upstream has found and fixed this bug in a new version. I'll try to take
a look at backporting the fix to the breezy version, if that would help.

Sebastien Bacher (seb128) wrote :

as mentionned with an another comment the fix has some crasher. I've tried to
combine it with some FC patch but there it doesn't work fine neither

Roland Dreier (roland.dreier) wrote :

Yes, I see... a naive backport of the obvious changes doesn't work.

Has anyone tried contacting Daniel Veillard and/or John McCutchan about a
backported fix for 0.1.5?

Roland Dreier (roland.dreier) wrote :

(In reply to comment 13)

I still seem to have gam_server leaking, even when booting with noinotify.
Right now, I have:

    $ cat /proc/cmdline
    root=/dev/sda1 ro noexec=off console=tty0 quiet splash noinotify

but gam_server is still leaking. I just killed a multi-gigabyte process, and I
already have:

    $ ps aux|grep gam_server
    roland 13201 1.2 1.8 45636 38196 ? R 13:02 0:18
/usr/lib/gamin/gam_server

I'm not sure if this is connected, but in my .xsession-errors file, I have tons
of output like

    invalid length 24902
    invalid length 24902
    invalid length 24902
    invalid length 24902
    end from FAM server connection

until finally the file ends with:

    ...Too much output, ignoring rest...

Roland Dreier (roland.dreier) wrote :

About my previous comment -- just to be clear that I don't simply have a large
gam_server process, I now have:

    $ ps aux|grep gam_server && sleep 60 && ps aux|grep gam_server
    roland 13201 0.7 9.0 193088 185672 ? S 13:02 1:13
/usr/lib/gamin/gam_server
    roland 13201 0.7 9.0 194100 186728 ? S 13:02 1:13
/usr/lib/gamin/gam_server

ie gam_server is up near 200 MB and growing by ~1 MB/min.

I haven't rebooted or anything, so I am still running a kernel booted with
"noinotify"

If there's any debugging info that would be useful, let me know.

Sebastien Bacher (seb128) wrote :

*** Bug 25925 has been marked as a duplicate of this bug. ***

Matti Airas (mairas) wrote :

There already is a month-old version 0.1.7 in the upstream. I wonder are there
any plans to backport that to Breezy? I am really suffering from this bug, and
since my system is a completely standard Ubuntu 5.10 ("any delta is a cost"
applies to end-users as well), there must be plenty of entry-level Ubuntu users,
who don't know what is going on and really can't fix the problem by themselves.

Sebastien Bacher (seb128) wrote :

This is fixed with this upload for dapper:

gamin (0.1.7-2ubuntu1) dapper; urgency=low

  * Sync with Debian.
  * debian/control.in,
    debian/python2.{3,4}-gamin.install:
    - changed the python version

 -- Sebastien Bacher <email address hidden> Tue, 22 Nov 2005 16:30:59 +0100

This bug doesn't happen to everybody (I've no issue by example). For the
backport I'm not sure, I'll ask to the backport team if they want to do this.

Sebastien Bacher (seb128) wrote :

*** Bug 26086 has been marked as a duplicate of this bug. ***

Sebastien Bacher (seb128) wrote :

*** Bug 26830 has been marked as a duplicate of this bug. ***

Michael R. Head (burner) wrote :

This bug has been hitting me hard this weekend. Previously, I'd only experienced
it once on one of my desktop machines. Now it's occured on my laptop 3 times
since Thursday.

Sebastien Bacher (seb128) wrote :

*** Bug 27236 has been marked as a duplicate of this bug. ***

Paweł T. Jochym (jochym) wrote :

(In reply to comment #24)
> This is fixed with this upload for dapper:
>
> gamin (0.1.7-2ubuntu1) dapper; urgency=low
>

I've done a backport of this to breezy (just apt-src thing)
and so far it looks fine. I'll report on this again in few days.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.