Re: vgchange (lvm) fails with large snapshots on low memory machines
>I am running 2.6.32-5-orion5x on a dns-323 (800MHz processor, 64MB
>When I have large snapshots (e.g., 75GB), I cannot activate the
>snapshots or the corresponding original lvm partition. In fact,
>running the command "vgchange -ay" (or the equivalent lvchange
>version) either hangs the machine or returns the line "killed" after a
>Looking in the syslog, I get many error messages about memory
>allocation failures, including:
> vgchange: page allocation failure. order:0, mode:0x10
>Nothing else is running on the machine and I have 500MB of swap
>(almost all free). I also have a reasonable (for this low memory
>machine) of unused RAM:
> total used free shared buffers cached
>Mem: 61056 51688 9368 0 340 7156
->/+ buffers/cache: 44192 16864
>Swap: 530104 6636 523468
>Everything works if I switch the drive to a plugcomputer with 512MB
>RAM though vgchange takes a couple of minutes to load (I believe that
>is because snapshots are slow).
>Also, it worked when I removed the older of the two 75GB snapshots.
>And I think it didn't become a problem until the snapshot got "old"
>(i.e., started to differ a lot from the original).
>So what is going on with vgchange that it cannot allocate enough
>kernel memory? Shouldn't the free RAM plus the swap be enough?
>Or is there something special about vgchange that requires actual RAM?
I am also having a problem on my 64meg of ram dns-323 when trying
rsync from drive to drive. I can rsync from a remove machine onto the
dns-323, but I recently rsynced 1600GB on lenny and had no problems
pre-upgrade to 6.0. Any help would be greatly appreciated as I love
---------- Forwarded message ----------
From: Martin Michlmayr <email@example.com>
Date: Mon, Feb 14, 2011 at 9:24 AM
Subject: Re: dns-323 upgrade form lenny to squeeze final
To: brossmeisl <firstname.lastname@example.org>
* brossmeisl <email@example.com> [2011-02-13 20:37]:
> First off, thank you so much for making it possible to install
> debian on the dns-323. I upgraded to squeeze when it went gold a week
> or so ago, and since then I am getting tons of oom-killer errors when
> I run certain programs (mainly rsync on directories locally). I was
> wondering if you had done anything with sysctl to tweak the kernel to
> be more kind to a systme with only 64megs of ram?
I haven't, but then again I don't use my DNS-323 anymore. It might be
worth asking on the debian-arm to see if other people see the same
problem. I would have assumed that 64 MB RAM should be enough but
then again I don't know what you're running on your DNS-323.