[gentoo-user] Re: Suspend to RAM caused crashes

[Francesco Talamona <francesco.talamona@know.eu>]

On Sunday 21 August 2011, Mick wrote:
> Here's a strange one:
> 
> Suspending a Pentium4 32bit machine used to work a treat.  For years.
>  Then around 9 months ago or so, I can't recall exactly, it started
> causing crashes. What happens is that the monitor will go to sleep
> and the disk will stop immediately, but the machine continues to run
> and run and run ...
> 
> At that point I have lost access to the keyboard and the monitor does
> not wake up if I move the mouse.  Using ssh to connect shows that
> the machine is off the network, so I assume that the NIC is also
> suspended.  The only way to recover is to pull the plug.  :-(
> 
> Unfortunately, mysql has left a lock file behind, so it won't start
> at reboot until I remove the lockfile.
> 
> Now, here's the strange thing about all this.  I have 4 RAM modules,
> 2x1G and 2x500M.  Following the manual I have installed them in this
> order:
> 
> slot 1 - 1G,
> slot 2 - 0.5G,
> slot 3 - 1G,
> slot 4 - 0.5G
> 
> If I try to suspend the machine soon after boot, when it is still
> using low amounts of memory, the machine will suspend each time
> without fail (just like it used to do in the past).
> 
> If I wait until the machine is using more than 1G or so, then it will
> always crash.
> 
> I'm running memtest86+ just in case, but 3 passes and no errors are
> shown so far.  Suspend to RAM is really a time saver on this machine
> and was being used at least 4-5 times a day.  Now the box is running
> non-stop 16 hours a day or more, which is wasteful (although with
> the Pentium4 I'm saving on central heating bills!)  Any ideas what I
> can look into to resolve this?

I'm using KDE to suspend my machine, for some reason it recently stopped 
working, but it was only with kernel 3.0.1

I was able to see an error quickly changing to tty12 where the console 
is used to echo /var/log/messages after clicking "sleep" button in KDE

Aug 15 07:42:31 aemaeth polkitd(authority=local): Registered 
Authentication Agent for unix-
session:/org/freedesktop/ConsoleKit/Session1 (system bus name :1.29 
[/usr/lib64/kde4/libexec/polkit-kde-authentication-agent-1], object path 
/org/kde/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8)
Aug 15 07:42:35 aemaeth dbus[3247]: [system] Rejected send message, 2 
matched rules; type="method_call", sender=":1.33" (uid=501 pid=4883 
comm="nautilus --sm-client-id 10c6d9d5610001307993940000") 
interface="org.freedesktop.DBus.Properties" member="GetAll" error 
name="(unset)" requested_reply="0" destination=":1.1" (uid=0 pid=3297 
comm="/usr/sbin/console-kit-daemon ")

The machine locked itself shortly after that error, but the log was 
there.

My problem is now gone with kernel 3.0.3. I wish yours it's not a RAM 
issue, it could be tricky to spot, because memtest is not putting any 
load to the machine, so it's very useful when it reports error, but when 
it doesn't you can't be sure if RAM modules are in good health.
From my experience the stress load of compiling large packages is more 
likely to evidence RAM faults than memtest itself.

HTH
	Francesco

-- 
Linux Version 3.0.0-gentoo, Compiled #3 SMP PREEMPT Fri Aug 5 21:02:22 
CEST 2011
Two 2.9GHz AMD Athlon 64 X2 Processors, 4GB RAM, 11659 Bogomips Total
aemaeth