[06:21] <joh6nn> i'm trying to add a script to my boot process to turn off dma on one of my hdds (long story), and i'm very clearly going about it the wrong way, but i'm not sure where i'm falling down
[06:21] <joh6nn> i've looked through the wiki and the getting-started page, but i'm guessing that's only taught me enough to be dangerous, not enough to get this working
[06:41] <wasabi__> /etc/rc.local =)
[06:43] <joh6nn> just stuff it in there and it should work?
[06:44] <wasabi__> uh huh
[06:44] <joh6nn> thanks, i'll give that a shot
[06:45] <joh6nn> alternatively, is there a better way to disable dma for a drive at boot?
[07:12] <joh6nn> er.  wasabi__, i must be missing something.  rc.local is a basically no-op script.  were you just suggesting i shove my one line in there, since it's there anyway?
[07:12] <wasabi__> uh huh
[07:13] <wasabi__> it's run during boot. it's for local stuff... to run during boot
[07:13] <wasabi__> this has nothing to do with upstart, btw. =)
[07:14] <joh6nn> yeah, normally, that'd be a fair enough answer to the question.  unfortunately, i'm needing to set /dev/hdd to dma = off round about the time that the init-top scripts start getting run (possibly sooner; i may be mis-remembering at that point)
[07:15] <wasabi__> that would be in the initramfs
[07:16] <joh6nn> yeah, that was where i tried to add it first.  i stuffed my own little script in there in init-top, but there's seemingly more to it than just adding new scripts to the directory; it didn't appear to run
[07:17] <wasabi__> have to rebuild hte initramfs
[07:17] <wasabi__> dpkg-reconfigure the kernel package
[07:18] <wasabi__> also, i dont think your drives will exists in init-top
[07:18] <wasabi__> maybe. ;)
[07:18] <joh6nn> that can't be right; i need to reconfigure the kernel package just to add a new boot script?
[07:20] <wasabi__> uh huh
[07:20] <wasabi__> that's hte initramfs
[07:20] <wasabi__> initramfs != boot scripts exactly.
[07:22] <joh6nn> ok, well, then i'm clearly going about this the wrong way.  there's gotta be a better way of changing this setting then
[07:23] <wasabi__> i'd like to know why you have to change it
[07:24] <joh6nn> if it i don't, it hangs and doesn't finish booting.  at a guess, the BIOS in the box isn't very good, and isn't properly handling the dma settings for the drive
[07:24] <wasabi__> hmm.
[07:25] <wasabi__> well, how far does the boot get?
[07:25] <joh6nn> gets to scanning to see what IDE devices are connected, and hangs at /dev/hdd
[07:26] <joh6nn> i've tried setting it in /etc/hdparm.conf, too, but that doesn't have any visible effect
[07:26] <wasabi__> It's going to be a bit hard to tell the kernel not to use dma on a device before the kernel hsa discovered hte device.
[07:29] <joh6nn> yeah, i'm grasping at straws, really
[07:29] <wasabi__> also, im not sure if the kernel even enables dma automatically in the initramfs or before
[07:29] <joh6nn> i've tried setting it with hdparm, and then rebooting, but that doesn't work.  i'm guessing it's not working because the BIOS sees that the drive supports DMA, and turns it back on
[07:30] <wasabi__> hdparm affects transient kernel settings.
[07:30] <wasabi__> it doesn't save
[07:30] <joh6nn> ah
[07:31] <joh6nn> would be nice if the man made that clearer, eh? ; )
[07:33] <wasabi__> well, anything that "saved" in linux would have to have someplace to 'save it'
[07:33] <wasabi__> hdparm.conf is that
[07:35] <wasabi__> off to bed. sorry i can't help ya much
[07:35] <wasabi__> night
[12:34] <Keybuk> random thought
[12:34] <Keybuk> the job definition parser currently "ignores" parser errors and just outputs warnings
[12:34] <Keybuk> should it abort parsing instead?
[12:34] <AlexExtreme> I would think so
[12:35] <AlexExtreme> because if there is an error, it could cause the rest of it to parse incorrectly and could cause problems
[12:35] <Keybuk> yeah, especially the unknown stanza bit ...
[12:35] <Keybuk> ned script
[12:35] <Keybuk>   exec foo
[12:35] <Keybuk> end block
[12:35] <Keybuk> will give you
[12:35] <Keybuk> foo:1: ignored unknown stanza: ned
[12:35] <Keybuk> foo:2: ignored unknown stanza: end
[12:36] <Keybuk> but set the job's exec to be "foo" because exec is a valid stanza
[12:36] <AlexExtreme> yep