[06:28] <fabbione> morning
[12:37] <Mithrandir> hiya
[12:37] <Mithrandir> on a freshly-installed hoary box, I get a loadavg of 6.0 with a set of kernel threads stuck in D
[12:37] <Mithrandir> : tfheen@rho / > ps ax | grep D PID TTY      STAT   TIME COMMAND 139 ?        D      0:00 [khubd]  182 ?        D      0:00 [kswapd1]  183 ?        D      0:00 [kswapd0]  774 ?        D      0:00 [kseriod]  953 ?        D      0:00 [md1_raid1]  954 ?        D      0:00 [md1_resync] 
[12:37] <Mithrandir> argh, that was totally unreadable
[12:38] <Mithrandir> anyhow, khubd, kswapd1, kswapd0, kseriod, md1_raid1, md1_resync.
[12:38] <Mithrandir> anybody got any bright ideas why they're stuck and how I can get them to unstick themselves?  Rebooting doesn't seem to help
[12:52] <fabbione> Mithrandir: what arch is that?
[12:53] <fabbione> it looks like there are disk I/O problems and the raid rsync dies
[12:53] <Mithrandir> amd64
[12:53] <fabbione> raid sync
[12:53] <Mithrandir> /proc/mdstat seems happy
[12:53] <Mithrandir> hmm
[12:53] <fabbione>              D      0:00 [md1_resync] 
[12:53] <Mithrandir> raid1: raid set md0 active with 2 out of 2 mirrors
[12:53] <Mithrandir> hdd: packet command error: status=0x51 { DriveReady SeekComplete Error }
[12:53] <Mithrandir> hdd: packet command error: error=0x54
[12:53] <fabbione> that can be a reason
[12:53] <fabbione> try to disable DMA, make it boot
[12:53] <fabbione> and see if it works
[12:54] <Mithrandir> hm, that's the CD-ROM. :-P
[12:54] <Mithrandir> this is a pure SATA box
[12:55] <fabbione> hmmmm
[12:55] <fabbione> can try using the SATA compatibility mode in the BIOS?
[12:56] <Mithrandir> it doesn't have one, afaik.
[12:56] <Mithrandir> this is a server, not a toy. :-)
[12:56] <fabbione> a real server uses SCSI
[12:56] <fabbione> not SATA :P
[12:56] <Mithrandir> http://err.no/tmp/dmesg is the dmesg
[12:56] <Mithrandir> SATA is the new SCSI
[12:57] <fabbione> a real server uses OLD and WORKING SCSI
[12:57] <fabbione> devfs_mk_dir: invalid argument.<4>devfs_mk_dev: could not append to parent for /disc
[12:57] <fabbione> wtf
[12:57] <Mithrandir> it uses LVM too.
[12:59] <fabbione> Stopping tasks: [12:59] <fabbione> ==<6>md: checkpointing recovery of md1.
[12:59] <fabbione> =
[12:59] <Mithrandir> hmm
[01:00] <Mithrandir> so twiddling to get around that should fix it?
[01:00] <fabbione> what controller is that?
[01:00] <Mithrandir> sii3114
[01:00] <fabbione> i am not sure.. but clearly the devfs_mk stuff is not normal
[01:01] <Mithrandir> fwiw, I see the same on a similar box which does not have the problem with load.
[01:02] <Mithrandir> hmm, I'll try rebooting it now that the raid is synced and such.
[01:09] <Mithrandir> ok, so _something_ funks up the resync
[01:09] <Mithrandir> I'm not sure what, but the reboot made it resync one of the raids
[01:20] <fabbione> is it still stucked?
[01:21] <Mithrandir> nope, but it resyncs the raid now.
[01:22] <Mithrandir> if you look at the previous dmesg, it does:
[01:22] <Mithrandir> Stopping tasks: [01:22] <Mithrandir> ==<6>md: checkpointing recovery of md1.
[01:22] <Mithrandir> = stopping tasks failed (1 tasks remaining)
[01:24] <fabbione> yup
[01:25] <Mithrandir> while the last restart worked.
[01:37] <fabbione> hey jbailey
[01:38] <jbailey> Heya Fabio!
[01:38] <fabbione> https://bugzilla.ubuntu.com/show_bug.cgi?id=12040
[01:39] <fabbione> jbailey: unfortunatly it is an initrd-tools problem
[01:39] <fabbione> i am going to add info on it, but we need it fixed asap
[01:43] <jbailey> fabbione: I don't have enough info here to see what the problem is....
[01:43] <jbailey> Is it the usual round of glibc fuckage?
[01:43] <fabbione> i just added them :)
[01:43] <fabbione> nope..
[01:44] <jbailey> Sure, this is just another driver alias thing.
[01:45] <fabbione> jbailey: remember that the megaraid still remains..
[01:45] <fabbione> the driver has been splitted in 2
[01:45] <fabbione> old and new...
[01:45] <fabbione> but i dunno how to fix that in mkinitrd
[01:46] <jbailey> There's a massive sed table in /usr/sbin/mkinitrd
[01:50] <Mithrandir> fabbione: I need to see if I can reproduce this problem, but it went away after the fully-synced system had booted.
[01:50] <fabbione> jbailey: yes, but this is not a rename or an alias
[01:50] <fabbione> Mithrandir: it's rather strange because i do usually test installs on raid
[01:50] <Mithrandir> fabbione: and LVM?
[01:50] <fabbione> yup
[01:50] <jbailey> fabbione: Oh, sorry, I see.
[01:51] <Mithrandir> fabbione: hmm, anyhow, I'll see if I can reproduce it
[01:51] <fabbione> Mithrandir: i do all kind of stuff here :P
[01:51] <jbailey> Sorry, I thought you mean megaraid for old kernels and megaraid_mbox for new ones.
[01:51] <fabbione> Mithrandir: lvm is not related to that
[01:51] <jbailey> I'm still sleepy. =)
[01:51] <fabbione> jbailey: ehehe no
[01:51] <fabbione> we splitted the driver
[01:51] <fabbione> so that all the cool stuff is from the new one
[01:51] <fabbione> the only 2 old controllers are supported by the old one
[01:51] <fabbione> (since the new one doesn't)
[01:52] <jbailey> Should I just modprobe the new one first and the old one second?
[01:52] <fabbione> jbailey: that would be fine.. yes
[01:53] <fabbione> we also have a test box...
[01:53] <jbailey> Lovely.
[01:53] <fabbione> so if you want we can test the fix
[01:53] <jbailey> When I've cleared the cobwebs out of my head, I'll make a test package.
[01:53] <jbailey> Was up very late for a city-wide party last night.
[01:53] <fabbione> ehhe
[01:54] <fabbione> it's enough you give me the modified mkinitrd file
[01:54] <fabbione> we can package it later :)
[03:16] <zul> jeua
[03:16] <zul> doh...heya
[03:19] <zul> fabbione: when you get around to it can you do a merge kthxbye
[03:24] <fabbione> zul: fix your archive. kthxbye
[03:24] <fabbione> baz abrowse zulcss@gmail.com--2005
[03:24] <fabbione> ^^stall
[04:06] <zul> fudge
[09:59] <dilinger> i have serious slowdowns switching between workspaces in hoary's X.  not sure if this is xorg or kernel being crap :(
[11:35] <dilinger> fabbione: ubuntu-kernel list
[11:35] <dilinger> Subject:      [ghall@research.dfci.harvard.edu: Re: qlogic] 
[11:37] <fabbione> dilinger: yes.. we just started compiling it again
[11:37] <fabbione> it's in the vanilla kernel
[11:37] <dilinger> fabbione: it's not actually distributable
[11:37] <fabbione> it might not be DFSG free...
[11:37] <dilinger> http://wiki.debian.net/?KernelFirmwareLicensing
[11:38] <dilinger> (i need to update that wrt to qlogic, but i've talked to their lawyers a bunch, and they're going to dual license the firmware.. gpl/bsd)
[11:39] <fabbione> dilinger: well the point is simple
[11:39] <fabbione> there are some kernel developers that are making real crusades towards non GPL code
[11:39] <fabbione> so if it is still there, i think it's usable
[11:39] <dilinger> no
[11:39] <dilinger> this is a different matter
[11:39] <dilinger> this is GPL'd firmware
[11:40] <dilinger> the GPL defines source as "preferred form for modification"
[11:40] <fabbione> is the firware within the kernel?
[11:40] <dilinger> which binary firmware is not
[11:40] <dilinger> yes
[11:40] <dilinger> qla2300_fw.c
[11:40] <dilinger> and friends
[11:40] <dilinger> that's why we strip it out of the kernel
[11:40] <dilinger> along w/ tg3
[11:40] <dilinger> and a few other drivers
[11:40] <dilinger> tg3 is cleared up in 2.6.12; we worked w/ broadcom
[11:41] <dilinger> we're working w/ qlogic, and qla2xxx should be good to go soon
[11:41] <dilinger> the other 6 drivers we strip are looking hopeless
[11:41] <dilinger> but they
[11:41] <fabbione> well i am already shipping them..
[11:41] <dilinger> 're not as widely used
[11:41] <fabbione> tg3/qla
[11:41] <dilinger> ok, well..
[11:41] <fabbione> also the drm stuff
[11:42] <dilinger> drm?
[11:42] <fabbione> mga (drivers/char/drm):
[11:42] <fabbione> ^^drm
[11:42] <dilinger> ok, that's fine
[11:42] <dilinger> that's a different issue
[11:43] <dilinger> DFSG-compatibility is the reason why that's listed there
[11:43] <fabbione> anyway.. i will look at it again another day
[11:43] <fabbione> i am way too tired to think about licence and firmwares
[11:43] <dilinger> this isn't about DFSG-ness, this is about undistributable code
[11:43] <dilinger> ok
[11:43] <fabbione> they should all just die
[11:43] <dilinger> 'night
[11:43] <dilinger> heh
[11:43] <fabbione> :)