latest F12 quckly unusable after restart

[Tom Diehl]

On Wed, 10 Feb 2010, Craig White wrote:

> On Wed, 2010-02-10 at 20:21 -0800, Ulrich Drepper wrote:
>> Since I updated my F12 workstation on Tuesday (haven't done it for two
>> weeks before that) my machine becomes unusable pretty quickly after
>> every restart.  Sometimes after 30 mins, sometimes after a few hours.
>>
>> It manifests itself in an ever increasing load, although top doesn't
>> show anything running.  Shell processes simply stop while typing.  I.e.,
>> they don't accept any more keystrokes and just freeze.
>>
>> X overall becomes very sluggish.  Scrolling in firefox moves pixels bn
>> lines.  The Xorg process is eating up lots of CPU cycles and kmsd (I
>> assume kernel mode switching support in the kernel) also shows up.
>>
>> At least before the last reboot I see tons of ext4-dio-unwrit processes.
>>   Last time 81 of them.  And I lost data at one of the many reboots.
>>
>>
>> The machine is a single socket i7 with plenty of RAM.  Lots of disk, one
>> RAID 0 and one RAID 1 among the filesystems.  All ext4.  The graphics
>> card is a dual head ATI.
>>
>> Anybody else seen something like that?
> ----
> I had 2 hard freezes in the last week but my setup is decidedly
> different. Using nouveau driver and nouveau.modesetting=0 as boot
> parameter and the first time, I thought it was the KVM I am using but
> the second time was over night / idle. I use KDE and I would guess that
> you are using Gnome.

Is the machine actually locked or just the X display. I have 2 new i7
machines using the nouveau driver and every couple of days the X display
locks. I can ssh to it from another machine and bring it down gracefully.
It needs to be powered off and restarted to get it back into a usable state.
If I just reboot it, tends to lock up before I can login. KVM seems to
aggravate the problem in that if I have virtual machines running the problem
happens sooner. IOW I get longer uptimes between lockups if the virt machines
are shutdown. I am not sure if that is relevant or not.

FWIW I filed Bug 559791 but so far no one has provided any feedback.

Regards,

-- 
Tom Diehl       tdiehl at rogueind.com      Spamtrap address mtd123 at rogueind.com