Ubuntu
gvfs package

[hardy][memory leak] gvfsd-smb uses >1GB memory after file copy

Bug #196991 reported by Sascha Grossenbacher
20
Affects Status Importance Assigned to Milestone
gvfs
Fix Released
High
gvfs (Ubuntu)
Fix Released
High
Unassigned

Bug Description

Binary package hint: gvfs-backends

Similar problem as in #5929

After copying some files from a smb:// share to my local disk, I noticed that the gvfsd-smb process does use 1,1GB memory, the same amount as the size of the copied files. It seems that all the copied files are in memory now and *stay* there. The copy command has successfully finished.

  PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
11139 berdir 20 0 1166m 1.1g 4148 S 0 54.3 0:15.02 /usr/lib/gvfs/gvfsd-smb --spawner :1.16 /org/gtk/gvfs/exec_spaw/4

Versions used:

ii gvfs 0.1.8svn200802 userspace virtual filesystem - server
ii gvfs-backends 0.1.8svn200802 userspace virtual filesystem - backends

I have added a screenshot of the "Speicherfelder" ( Don't know the english name, probably something like memory fields, right click on the process in the system overview and selected there "Speicherfelder")

At the moment, it isn't an actual problem, but the process will probably crash if one tries to copy more data than memory + swap

If I can give you more information, please tell me how :)

Related branches

lp:debian/squeeze/gvfs
Revision history for this message
Sascha Grossenbacher (berdir) wrote :
Revision history for this message
Nold (arnold-skimminge) wrote :

I have the same problem using ssh

Arnold

Revision history for this message
Oleksij Rempel (olerem) wrote :

I can confirm this issue.

Changed in gvfs:
status: New → Confirmed
Revision history for this message
Oleksij Rempel (olerem) wrote :
Revision history for this message
Oleksij Rempel (olerem) wrote :
Travis Watkins (amaranth)
Changed in gvfs:
importance: Undecided → High
status: Confirmed → Triaged
Bug Watch Updater (bug-watch-updater)
Changed in gvfs:
status: Unknown → New
Revision history for this message
Nikolaus Filus (nfilus) wrote :

same problem here, while copying local -> smb://

This problem should be marked critical as it occurs in a "daily" function and renders the system unusable after the memory usage reaches the memory maximum.
unmounting the automatic made smb:// share returns the occupied memory from gvfs-smb.

Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package gvfs - 0.1.8svn20080303-0ubuntu1

---------------
gvfs (0.1.8svn20080303-0ubuntu1) hardy; urgency=low

  * New svn snapshot:
    - correctly free the ressources after copying (lp: #196991)
    - no obex crash when mounting doesn't work correctly (lp: #197001)
    - no cdda crash when mounting doesn't work correctly (lp: #196692)

 -- Sebastien Bacher <email address hidden> Mon, 03 Mar 2008 18:44:24 +0100

Changed in gvfs:
status: Triaged → Fix Released
Bug Watch Updater (bug-watch-updater)
Changed in gvfs:
status: New → Fix Released
Revision history for this message
Charles Yates (charles-yates) wrote :

The issue still occurs with http copy (hardy gvfs 0.2.3-0ubuntu4) - gvfsd-http grows enormously when dragging and dropping files to desktop/nautilus.

Revision history for this message
Mika Fischer (zoop) wrote :

Confirmed on hardy final. And it's not a subtle leak. See attached screenshot.

Mika Fischer (zoop)
Changed in gvfs:
status: Fix Released → Confirmed
Revision history for this message
Sebastien Bacher (seb128) wrote : Re: [hardy][memory leak] gvfsd-http uses >1GB memory after file copy

don't reopen a closed bug because you have a similar issue, the bug was about the smb code and is a different topic than webdav

Changed in gvfs:
status: Confirmed → Fix Released
Revision history for this message
Mika Fischer (zoop) wrote :

If you had checked the upstream bug, this bug is not limited to SMB. Also this has nothing to do with webdav...

But fine, I'll open a new bug...

Revision history for this message
Sebastien Bacher (seb128) wrote :

did you read the upstream bug? the issue described there has been fixed, I think your problem is webdav specific and is due to the fact that datas copied are store in memory and not read in a standard way

Revision history for this message
Mika Fischer (zoop) wrote :

Of course I read the upstream bug. There it is claimed that job objects are not freed, which seems to be exactly the issue I'm having.
Yes, the issue is claimed to be fixed. But sometimes fixes are wrong/incomplete.

Anyway, I filed the new bug, let's continue discussion there.

But I really fail to see where you get the idea that webdav has anything to do with this. It was not mentioned in this bug by anyone but you...

Revision history for this message
Alwyn Schoeman (alwyn-schoeman) wrote :

I still get this bug using SMB on Karmic. gvfs-1.4.1-0ubuntu1

Revision history for this message
jhouse (jhouse) wrote :

I am experiencing this on Karmic as well.

Bug Watch Updater (bug-watch-updater)
Changed in gvfs:
importance: Unknown → High
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Duplicates of this bug

Subscribing...

Other bug subscribers

Remote bug watches

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