Cannot allocate memory

Bug #14847 reported by Jesir Vargas
4
Affects Status Importance Assigned to Milestone
sysvinit (Ubuntu)
High
Unassigned

Bug Description

After upgrading two machines from Warty to Hoary RC, and thus to kernel
2.6.10-686 and 2.6.10-k7, memory-allocation errors
have become very common.

The 686 is a P2-128MB.
The k7 is a Thunderbird-1GB.

The machines don't suffer from this with kernel 2.6.8 (the Warty one) and kernel
2.6.11 (the Hoary RC one).

The symptoms include low responsiveness and programs crashing (with a syslog
message of the type "Killing process: Out of memory", "fork: Cannot allocate
memory").

Revision history for this message
Fabio Massimo Di Nitto (fabbione) wrote :

What kind of applications are you running on these machines? did you setup a
swap partition?
if so how big it is?

Please attach the output of:

cat /proc/meminfo
ps aux
dmesg

This is mostlikely a process that is leaking memory. The kernel is doing the job
as it should.

Fabio

Revision history for this message
Jesir Vargas (jvargas) wrote :

Created an attachment (id=1967)
Files requested by Fabio

Here are the requested files plus fstab (showing that there's swap).
The most "troublesome" (and kernel-killable) processes seem to be firefox-bin,
dpkg/apt (when dist-upgrading/installing) and Xorg.
The files document only the 128MB system, as it is the one that "suffers" the
most.
Thanks.

Revision history for this message
Jesir Vargas (jvargas) wrote :

More INFO (cf. attachment) added as per Fabio's request.

Revision history for this message
Fabio Massimo Di Nitto (fabbione) wrote :

Well ... the point is very simple..

SwapTotal: 0 kB
SwapFree: 0 kB

swap is not active at all and yes i can see that it is there.
In any case this isn't a kernel bug, because swap is enabled in userland.

Please try the following steps:

mkswap /dev/hda5 (please double verify with your partition table!)
swapon -a

If this work, please do a reboot and check if the swap is enabled
automatically (as it should be) by initscripts.

Fabio

PS reassigning the bug to initscripts.

Revision history for this message
Jesir Vargas (jvargas) wrote :

It now works after rebooting, thanks again.

Revision history for this message
Fabio Massimo Di Nitto (fabbione) wrote :

I am still not sure what caused the problem. Did you only reboot? or did you
have to do anything more than that?

Revision history for this message
Jesir Vargas (jvargas) wrote :

Only did what you suggested, nothing more nor less.

What amuses me is that, when tested with 2.6.11, for example, it didn't occur.

Revision history for this message
Matt Zimmerman (mdz) wrote :

Bug #13808, perhaps?

Revision history for this message
Jesir Vargas (jvargas) wrote :

Perhaps indeed.

Revision history for this message
Matt Zimmerman (mdz) wrote :

Can you confirm whether you attempted to hibernate the system (in particular if
the resume failed)?

Revision history for this message
Jesir Vargas (jvargas) wrote :

Indeed, on both machines, I tried the suspend-to-disk feature as outlined at

http://www.ubuntulinux.org/wiki/HoaryPM

I confirm each machine failed to "resume".

Revision history for this message
Fabio Massimo Di Nitto (fabbione) wrote :

This bug has been marked as a duplicate of bug 13808.

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.