| edgecase | last hope, oh my! | 00:40 |
|---|---|---|
| Keybuk | heh | 00:43 |
| edgecase | ok semantic question, on ubuntu mountall.conf takes care of mounting root fs rw, other fs, etc, | 00:43 |
| Keybuk | yes | 00:44 |
| edgecase | but, could it be assumed, that by definition, starting /sbin/init, means the rootfs is already mounted, possibly read-only? | 00:44 |
| Keybuk | yes | 00:45 |
| edgecase | i mean either a monolithic kernel, or an initrd's pivot_root, or say lxc container starting... | 00:45 |
| Keybuk | /* Sanity check, the root filesystem should be already mounted */ | 00:45 |
| Keybuk | root = find_mount ("/"); | 00:45 |
| Keybuk | if (! root->mounted) { | 00:45 |
| Keybuk | nih_fatal ("%s", _("root filesystem isn't mounted")); | 00:45 |
| Keybuk | exit (EXIT_ERROR); | 00:45 |
| Keybuk | } | 00:45 |
| Keybuk | even mountall assumes that :p | 00:45 |
| edgecase | so "start on startup" implies certain parts of FHS are available, yes? | 00:45 |
| Keybuk | edgecase: it implies that the root filesystem is read-only | 00:45 |
| Keybuk | we tend to assume that means you have /bin, /sbin, /lib and /etc | 00:46 |
| edgecase | but not /usr or /var | 00:46 |
| Keybuk | right | 00:46 |
| Keybuk | definitely not /usr, /var, /boot, /home, /proc, /sys, /dev, etc. | 00:46 |
| Keybuk | although with Upstart, you can assume /proc and /sys are mounted | 00:46 |
| edgecase | i'm just pondering the sysctl.conf race problem | 00:47 |
| Keybuk | and with recent kernels, /dev should be a devtmpfs | 00:47 |
| Keybuk | sysctl.conf race problem? | 00:47 |
| edgecase | race of setting things vs network interfaces coming up. in Karmic for example | 00:47 |
| edgecase | there's a couple launchpad bug IDs | 00:48 |
| edgecase | i drew out an elaborate map of the issues, but let me try and focus on one aspect | 00:48 |
| edgecase | ok, with sysctl parameters, there are some that affect parts of the kernel which is only compiled in (not a module) | 00:50 |
| edgecase | but with many, it could be a module or compiled in, so there's no *right* place for sysctl.conf to be loaded | 00:50 |
| Keybuk | right | 00:52 |
| Keybuk | this is why I keep trying to eradicate that silly directory | 00:53 |
| Keybuk | but people keep insisting I put it back :p | 00:53 |
| edgecase | i wonder if the whole sysctl paradigm doesn't work so well | 00:53 |
| Keybuk | I tend to think that, since we have the kernel source, changing a sysctl default is best done by changing the default in the kernel | 00:54 |
| Keybuk | otherwise you have races like "the sysctl doesn't get set until after ssh is running and there's a security hole for a while" | 00:54 |
| edgecase | i'm thinking of network configs here, default isn't enough | 00:54 |
| edgecase | the system needs to be configurable | 00:54 |
| Keybuk | sure | 00:55 |
| Keybuk | but you can do one off-things like that with udev | 00:55 |
| edgecase | there's a couple real problems with sysctl tho, | 00:55 |
| Keybuk | SUBSYSTEM=="module", KERNEL=="i915", RUN+="echo 4 > /proc/sys/..." | 00:56 |
| edgecase | a NIC module could be compiled in, loaded by udev, or by the initrd... hard to put it's settings in one place had have it affect all 3 | 00:56 |
| Keybuk | you should still get module events for them | 00:57 |
| edgecase | will that read sysctl.conf ? | 00:57 |
| Keybuk | you only don't get module events for built-ins with no parameters | 00:57 |
| Keybuk | no | 00:57 |
| edgecase | hence the sticky issue | 00:57 |
| Keybuk | and reading sysctl.conf 800 times during boot is out of the question, before you ask ;) | 00:57 |
| edgecase | workaround for the parameters i need is to set in /etc/network/interfaces and avoid the issue | 00:58 |
| edgecase | but i'm also interested in the bigger picture of upstart | 00:58 |
| edgecase | maybe just agree that sysctl isn't good for things not compiled-in | 00:59 |
| Keybuk | I don't think there's a picture of sysctl in the upstart painting yet | 01:00 |
| Keybuk | it's probably in the big white space off to the right with a doodle of someone's dog in it | 01:00 |
| Keybuk | sorry, have to run now | 01:01 |
| Keybuk | it's 1am here | 01:02 |
| edgecase | yeah, that's ok, i can workaround for the network stuff i need | 01:02 |
| Keybuk | feel free to mail suggestions for how this could work to the ML | 01:03 |
| Keybuk | upstart is very much still under development | 01:03 |
| Keybuk | so lots of things can be improved | 01:03 |
| edgecase | d'oh! | 01:03 |
| edgecase | be around tomorrow? | 01:03 |
| Keybuk | yes, though earlier! | 01:04 |
| Keybuk | nite | 01:04 |
| edgecase | ok | 01:04 |
| edgecase | thx | 01:04 |
| edgecase | hmm hostname.conf, nothing depends on it, | 01:15 |
| NoReflex | hello! I'm having some problems with upstart in karmic server 64bit. It won't start postgresql. I tried http://superuser.com/questions/98702/how-to-make-postgresql-start-at-boot-time-in-ubuntu but it didn't work. can I remove upstart? I'm afraid to test it because the machine is at a remote location and if I mess it up I won't be able to connect to it anymore | 07:38 |
| NoReflex | I can start postgresql using /etc/init.d/postgresql-8.4 manually after login. There are links in /etc/rc*.d to postgresql init script but for some reason it isn't started automatically after reboot | 07:38 |
| NoReflex | nothing? | 08:21 |
| elpargo | hi. I got a program that I want to "always be running" as it opens a http listener and waits for input "forever" | 08:53 |
| elpargo | I'm not entirely sure if I should be doing an upstart script for it. | 08:53 |
| ct529 | hi everybody .... I am having serious problems with upstart on ubuntu 910 64 bit .... it would not start CUPS at all .... | 09:54 |
| ct529 | CUPS is properly configrued in /etc/rcX.d and in /etc/init.d .... | 09:54 |
| NoReflex | ct529: I have the same problem with upstart on Karmic 64bit. On my machine it won't start postgresql server. I tried http://superuser.com/questions/98702/how-to-make-postgresql-start-at-boot-time-in-ubuntu without luck | 10:07 |
| NoReflex | ct529: as a workaround you can use rc.local ... not perfect but at least it works... | 10:09 |
| ct529 | NoReflex: how? | 10:11 |
| ct529 | NoReflex: yes, I have the same problem with mysql and postgresql | 10:11 |
| ct529 | NoReflex: what do you see when you run the command runlevel? | 10:12 |
| NoReflex | ct529: what I did was to remove the links from /etc/rc*.d for postgres (sudo update-rc.d -f postgresql-8.4 remove) and add the start command to rc.local | 10:13 |
| ct529 | NoReflex: did you add it manually? | 10:13 |
| ct529 | NoReflex: could you pleas elet me know what do you see when you run the command "runlevel"? | 10:13 |
| NoReflex | ct529: runlevel gives N 2 | 10:14 |
| ct529 | NoReflex: so it works on your machine | 10:14 |
| edgecase | elpargo, are you asking if you should write upstart script instead of sysVinit script? | 14:05 |
| kimju | hello.. Is there some documentation on howto debug the bootup sequence? I'm getting bitten by bind9 trying to start (and failing) before the interfaces with ip-addresses that it is configured to listen-on / listen-on-v6 are brought up. | 14:17 |
| kimju | (this is happening on ubuntu karmic. someone else has reported it as bug: https://bugs.launchpad.net/ubuntu/+source/bind9/+bug/510587 ) | 14:19 |
| Keybuk | kimju: you should ask on #ubuntu about that | 14:21 |
| Keybuk | as far as I'm aware, bind9 in ubuntu is still using a sysvinit script and hasn't been converted to an upstart config file | 14:21 |
| kimju | Keybuk, ok, sorry. I just asked before you joined the channel if there is some documentation on how to debug the bootup sequence, as I'm getting bitten by that bug. | 14:22 |
| Keybuk | boot with --verbose on the kernel command-line | 14:23 |
| edgecase | Keybuk, top o' the morning to ya! | 14:34 |
| edgecase | hmm hostname.conf, nothing depends on it, | 14:35 |
| edgecase | so a race is possible with anything that uses hostname/utsname | 14:35 |
| Keybuk | edgecase: hey | 14:37 |
| Keybuk | ah yes, it's still St Patrick's Month | 14:37 |
| edgecase | i will now continue the barrage! | 14:37 |
| Keybuk | edgecase: can you think of anything that cares about the hostname? :p | 14:37 |
| edgecase | syslog? | 14:37 |
| Keybuk | it doesn't really depend on it | 14:38 |
| Keybuk | it queries it for every log | 14:38 |
| edgecase | ah ok | 14:38 |
| Keybuk | so if the hostname changes, it works just fine | 14:38 |
| edgecase | aside, is there anything to graph the dependencies visually? | 14:38 |
| edgecase | would make investigation easier | 14:39 |
| edgecase | for rsyslog, if you log to remote host, might get unknown hostname for a bit? | 14:39 |
| Keybuk | no, but there's no reason it shouldn't be easy to write something that makes dotty output | 14:40 |
| edgecase | ok will look at dotty | 14:40 |
| Keybuk | for rsyslog, if you log to remote host, you have to wait for the network to come up first :p | 14:40 |
| Keybuk | and on the typical laptop system, that won't happen until a user logs in and unlocks the WPA key :p | 14:40 |
| edgecase | you noticed my nick is "edgecase" right? | 14:40 |
| edgecase | i tend to ignore "typical" cases :P | 14:41 |
| Keybuk | :) | 14:41 |
| edgecase | anyhow i'll admit hostname is a trivial case | 14:41 |
| edgecase | i notice mountall.conf emits many variations of filesystems | 14:42 |
| edgecase | is there a formal definition of those, that would allow splitting them up? | 14:42 |
| edgecase | "emits filesystem", does that mean root fs is read-write? | 14:42 |
| edgecase | appologies if that info is published already | 14:43 |
| edgecase | hmm, having /var not a mandatory part of root fs, and yet having /var/run mounted on top, seems problematic | 14:50 |
| edgecase | I question the original unix paradigm of "map everything to a file" for daemon .pid files, esp. if they are going to just map back to ram(disk) | 14:51 |
| edgecase | just use unix socket or even abstract namespace one | 14:52 |
| Keybuk | edgecase: did you do "man 7 filesystem" ? | 14:54 |
| Keybuk | problematic> not really | 14:54 |
| Keybuk | it requires some juggling | 14:54 |
| Keybuk | but it works | 14:54 |
| Keybuk | we have /var/run before and after /var is mounted | 14:54 |
| Keybuk | in fact, even as /var is mounting, /var/run is still present | 14:54 |
| edgecase | ok, juggling is acceptable if you don't drop the ball | 14:54 |
| Keybuk | *shrug* | 14:55 |
| Keybuk | if the boot sequence drops any ball, you can't boot | 14:55 |
| edgecase | badness++ | 14:55 |
| Keybuk | why? | 14:56 |
| Keybuk | just write things to be reliable | 14:56 |
| edgecase | i'll just assume "even as /var is mounting, /var/run is still present" is relable, moving on... | 14:57 |
| Keybuk | yes | 14:57 |
| Keybuk | I could teach a course on Theory of Atomic Operations in Operating Systems 101 :p | 14:57 |
| Keybuk | I think about them a lot | 14:57 |
| Keybuk | the details: | 14:57 |
| edgecase | oh thank god | 14:57 |
| Keybuk | we always have a /var/run directory on the root filesystem, under the /var directory | 14:58 |
| Keybuk | we mount /var/run there on the first pass | 14:58 |
| Keybuk | then when it's time to mount /var, we actually mount it somewhere else (/dev/.var I think :P) | 14:58 |
| edgecase | is there a mount --under option or something? | 14:58 |
| Keybuk | we bind mount the current /var/run to /dev/.var/run | 14:58 |
| Keybuk | so now it's available at both /var/run and /dev/.var/run | 14:58 |
| Keybuk | we then move the /dev/.var mount to /var | 14:59 |
| edgecase | all atomic? | 14:59 |
| Keybuk | each of those is atomic yes | 14:59 |
| Keybuk | since a mount() operation holds the VFS lock | 14:59 |
| edgecase | ooh, snap! | 14:59 |
| Keybuk | mountinfo will show /var/run mounted twice | 14:59 |
| Keybuk | with different parent mounts | 14:59 |
| Keybuk | if you unmount /var, for whatever reason, /var/run is still mounted under it | 15:00 |
| edgecase | i find no mountinfo or man 7 filesystem | 15:00 |
| Keybuk | edgecase: are you on lucid? | 15:00 |
| edgecase | karmic | 15:00 |
| edgecase | i guess i need to be bleeding edge to get bleeding edge | 15:00 |
| Keybuk | ah, I think I added that manpage in lucid | 15:00 |
| edgecase | i'll get lucid on a box shortly | 15:00 |
| edgecase | so i can discuss current state of things | 15:01 |
| Keybuk | mountinfo => /proc/self/mountinfo | 15:01 |
| edgecase | ah /proc/mounts on steroids | 15:01 |
| edgecase | ok so, what provides /proc and /sys, upstart, before running any scripts? | 15:02 |
| Keybuk | I don't quite understand your question | 15:02 |
| Keybuk | you asked me what kind of trees apples, oranges and buses grow on :p | 15:02 |
| edgecase | earlier you said assume /proc and /sys when running upstart | 15:03 |
| Keybuk | yes | 15:03 |
| edgecase | so what mounts them? | 15:03 |
| Keybuk | init | 15:03 |
| edgecase | before running any "start on startup" jobs? | 15:03 |
| Keybuk | yup | 15:03 |
| edgecase | ok just getting my bearings | 15:03 |
| Keybuk | Upstart uses /proc/self/fd to run shell scripts | 15:03 |
| Keybuk | so it needs /proc mounted | 15:04 |
| Keybuk | and since it's probably going to need /sys at some point soon, I figured I'd mount that too | 15:04 |
| Keybuk | /dev gets mounted by the kernel these days | 15:04 |
| edgecase | is there a formal definition of state when "startup" event fires? | 15:04 |
| edgecase | ie "you get root /proc and /sys" | 15:04 |
| edgecase | and "but don't assume /var and /usr etc" | 15:05 |
| Keybuk | no written down one I guess | 15:05 |
| edgecase | use the source luke | 15:05 |
| Keybuk | the root filesystem is read-only | 15:05 |
| Keybuk | you have /proc and /sys | 15:05 |
| Keybuk | you probably have /dev | 15:05 |
| edgecase | reading source takes time tho | 15:05 |
| Keybuk | you have /bin, /sbin, /lib and /etc | 15:05 |
| Keybuk | but nothing else filesystem wise | 15:05 |
| edgecase | aside, i'm thinking mountall should be broken up, it's a syncronisation point, which defeats some parallelism | 15:06 |
| Keybuk | it isn't a synchronisation point | 15:06 |
| Keybuk | that's kinda the idea | 15:06 |
| edgecase | make a dummy job with the same syncronisation properties for dumb legacy jobs, | 15:06 |
| Keybuk | it replaces the shell script nightmares we had before that were | 15:06 |
| edgecase | but let smart jobs pick off say just root fs being rw | 15:07 |
| Keybuk | they can | 15:07 |
| Keybuk | on mounted MOUNTPOINT=/ | 15:07 |
| edgecase | mountall emits events? | 15:07 |
| edgecase | oh vfs emits mount events i'm guessing | 15:07 |
| Keybuk | no, mountall emits them | 15:08 |
| edgecase | back to upstart? | 15:08 |
| Keybuk | yes | 15:08 |
| Keybuk | like filesystem for a start ;) | 15:08 |
| edgecase | ok, then back to all the emits in mountall.conf... | 15:08 |
| edgecase | any definition what they mean? | 15:09 |
| Keybuk | emits virtual-filesystems | 15:09 |
| Keybuk | emits local-filesystems | 15:09 |
| Keybuk | emits remote-filesystems | 15:09 |
| Keybuk | emits all-swaps | 15:09 |
| Keybuk | emits filesystem | 15:09 |
| Keybuk | emits mounting | 15:09 |
| Keybuk | emits mounted | 15:09 |
| Keybuk | (from lucid) | 15:09 |
| Keybuk | for any of those, "man 7 EVENT-NAME" | 15:09 |
| Keybuk | (in lucid) | 15:09 |
| edgecase | ok will look there | 15:09 |
| Keybuk | I've tried to be reasonable well behaved about writing man pages for events | 15:09 |
| edgecase | that's why i say sync point, they all get emitted in parallel | 15:09 |
| Keybuk | they don't | 15:09 |
| edgecase | but only after *all* of mountall is done | 15:09 |
| Keybuk | no they don't | 15:09 |
| Keybuk | they get emitted at completely different times | 15:10 |
| edgecase | ok so it's a declaration, "may emit..." | 15:10 |
| Keybuk | yes | 15:10 |
| edgecase | right-o | 15:10 |
| Keybuk | did you read "man 5 init" ? :) | 15:10 |
| Keybuk | emits EVENT... | 15:10 |
| Keybuk | All processes on the system are free to emit their own events by | 15:10 |
| Keybuk | using the initctl(8) tool, or by communicating directly with the | 15:10 |
| Keybuk | init(8) daemon. | 15:10 |
| Keybuk | This stanza allows a job to document in its job configuration | 15:10 |
| Keybuk | what events it emits itself, and may be useful for graphing pos‐ | 15:10 |
| edgecase | i see now that i'm way to pessimistic coming into this | 15:10 |
| Keybuk | sible transitions. | 15:10 |
| Keybuk | emits is basically there to make writing that graphing tool you were asking about possible ;) | 15:11 |
| Keybuk | documentation only | 15:11 |
| edgecase | ok things look in order then | 15:11 |
| edgecase | i'm thinking then that just the .conf files can be optimized, to depend only on what they *really* need, | 15:12 |
| edgecase | to enable some more unusual use-cases than "user logs in then unlocks WLAN key..." | 15:12 |
| Keybuk | yes, certainly | 15:12 |
| edgecase | like, network booting etc | 15:12 |
| Keybuk | the difficulty comes in adapting to users | 15:12 |
| Keybuk | like we could write the job to actually depend on /usr rather than filesystem | 15:12 |
| edgecase | exactly | 15:13 |
| Keybuk | but that wouldn't work if a user decided to put a different filesystem for /usr/bin and for /usr/lib | 15:13 |
| edgecase | FHS doesn't allow that does it? | 15:13 |
| === elpargo_ is now known as elpargo | ||
| Keybuk | yes it does | 15:13 |
| Keybuk | FHS only says directories you shouldn't put on different filesystems | 15:13 |
| edgecase | ok well on that topic, | 15:13 |
| Keybuk | things like /usr and /var are free-for-alls | 15:13 |
| edgecase | windows has "reparse points"... | 15:13 |
| Keybuk | it's pretty common for example to have /var/lib or /var/spool on different drives | 15:13 |
| edgecase | true | 15:14 |
| edgecase | with unix, a (empty) directory *could* be a mount point, no way to know | 15:14 |
| edgecase | you can read fstab, and hope it's in sync with reality | 15:14 |
| Keybuk | reparse points => those are basically crazy symlinks iirc? | 15:15 |
| edgecase | takes place of fstab entries | 15:15 |
| edgecase | you get file dir symlink, and now reparse point fs object | 15:15 |
| edgecase | so, it's like an empty dir, because it has a name, | 15:15 |
| Keybuk | actually it might not even be in fstab | 15:15 |
| Keybuk | e.g. if you use LABEL to mean mountpoint | 15:15 |
| Keybuk | or if it's a removable device under /media | 15:15 |
| edgecase | but it's a type explicityly marking it as a mount point | 15:15 |
| edgecase | ok then that's an arguement for having reparse points in linux | 15:16 |
| edgecase | although 75% of the unix world will issue a death warrant to you | 15:16 |
| edgecase | for saying it | 15:16 |
| edgecase | the cases that are in fstab could be handled | 15:16 |
| edgecase | maybe that's a starting point | 15:17 |
| edgecase | but it's not /usr i'm interested in... but rather starting things that need root fs read-only and /var/run available, starting those asap | 15:19 |
| edgecase | anyhow i see that upstart is thankfully ahead of my expectations, i need to do some more reading now | 15:19 |
| edgecase | thanks for the good work! | 15:19 |
| Keybuk | local-filesystems should be ok for those on Ubuntu | 15:30 |
| Keybuk | ah | 15:30 |
| Keybuk | sorry | 15:30 |
| Keybuk | virtual-filesystems | 15:30 |
| Keybuk | virtual-filesystems means you have root read-only, and all the things like /var/run and stuff | 15:30 |
| Keybuk | we use it to start things like udev, dbus, etc. | 15:30 |
| edgecase | k thx | 15:46 |
| wasabi | oh hi/bye edgecase | 17:33 |
| === elpargo_ is now known as elpargo | ||
| === elpargo_ is now known as elpargo | ||
| halfline | sadmac: did you ever look at the display manager selection thing more? | 20:56 |
| halfline | i mean http://rstrode.fedorapeople.org/quest_to_find_a_dm.txt | 20:56 |
| === elpargo_ is now known as elpargo | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!