Arch Linux

whoops

Member

Registered: 2009-03-19

Posts: 891

kernel 2.6.30: 10 seconds wait + raid + pm-utils resume

Hi!

So, with the new kernel I've got:
- md0 wait 10 seconds message that slows my boot down by 10 seconds every time (problems with that have been mentioned in other threads not connected to raid/resume)
- pm-utils hibernate doesn't work that well: it just boots normally most of the time instead of resuming (some issues like that with pm-utils have been mentioned in other threads/bugs not connected to raid / 10 seconds warning)
- my md0 raid sometimes fails to be created at boot (similar issues have been posted, too, not connected to pm-utils or wait message in any way)

... if I got it right so far. Now:

My swap drive is md0, which is raid, so the three problems (reported independently from each others from others) are connected in my case and I'm confused. Maybe I'm experiencing only one ortwo of the issues mentioned in other threads... and bugs... and I'm not sure where.. what... origins. Or where to start.

Figuring out why raid isn't working as well as before? How...? I tried to write a bug report about this, but I've got no idea how to describe it in a way that makes sense. Is it mdadm... or the kernel... or pm-utils... no idea.

Any hints? thx!

broch

Banned

From: L.A. California

Registered: 2006-11-13

Posts: 975

Re: kernel 2.6.30: 10 seconds wait + raid + pm-utils resume

1) you can try to install second kernel (trouble free for your setup) in parallel
2) remove pm-utils

if previous kernel works then problem is related to the new kernel and maybe pm-utils, if you remove pm-utils and boot to the new kernel and problem persists, then issue is related to the new kernel, if removing pm-utils helps then combination of pm-utils with new kernel does not work.

You may then either wait for new 2.6.30 or updated pm-utils (after filling bug report).. or go back to 2.6.29 and wait for 2.6.31 which seems to have a lot of bugfixes for 2.6.30 and it is more exciting anyway.

hope this will help a little

whoops

Member

Registered: 2009-03-19

Posts: 891

Re: kernel 2.6.30: 10 seconds wait + raid + pm-utils resume

That unconfused me a little bit, thanks!

So... I tried around some more. It seems like it could be very related to that similar sounding problem with same kernel in testing I had a little while ago after all, but I'm not sure. I first thought, it was something entirely else with similar symptoms, as the testing-thread-problem was constant (now it works at least 50%) and didn't have the 10 seconds delay (The message was showing, but it was not actually waiting the whole 10 seconds but just 0.5 or something like that).

- removing pm-utils obviously helps, as only my swap md0 is affected (but I practically never used to reboot, only hibernate, as it takes me a while to set everything up again after reboot). It looks as if it wouldn't work during boot when it's needed to resume, but later it works without any problems.
- downgrading the kernel helps too (but that messes up my already upgraded / compiled virtualbox, bin32, nvidia etc. stuff.) and get me back the old behaviour
- defining a normal non-raid swap fixes the problem too, I think. At least it hibernated & resumed 4 times without any problems.

... so, I did a pretty good job reconfusing myself

whoops

Member

Registered: 2009-03-19

Posts: 891

Re: kernel 2.6.30: 10 seconds wait + raid + pm-utils resume

So, as I don't have time for systematic testing ATM, I used my fall-back "incredible windows repairing skillz(tm)" (=doing random stuff all at once until I think it's better) on the machine:

- took out bios battery, updated bios, reconfigured bios
- enabled testing repository & updated everything to testing
- removed the kernel parameters defining the mdadm partitions from the kernel line in grub / menu.lst
- added mdadm and lvm2 to the hooks in mkinitcpio
- restored mdadm.conf.pacnew > mdadm.cong and had mdadm >> the partition list into it again
- added "/dev/sd[abc][12345]" to the device list in mdadm.conf (although "partitions" was already enabled by default)
- reinstalled kernel, reinstalled grub, rewrote grub to both hd's
- disassembled, deleted, recreated, reassembled, formatted swap md0
(- also I blew into the hd's contacts, hit / shook them a bit and put them back in for good luck. )

Seems to work at least a bit better now, gonna / gotta try hibernating 10-20 times to be sure though (can't atm). Prior to the kernel update, my hibernate was about 99% safe with the old config (never rebooted), now the first 4 times went fine again and I hope it will stay that way. Going to slowly revert the config changes after that, maybe I'll figure out what helped like that.

(so: almost solved maybe, gonna see...)

Last edited by whoops (2009-06-27 12:05:46)