**** BEGIN LOGGING AT Fri Apr 08 23:59:56 2005 Apr 09 01:01:55 * jbowler is away: bbl Apr 09 01:25:55 jbowler: re uClibc stat-bug... Does that mean it's not going to get fixed? Apr 09 01:29:53 rwhitby-away: By the way, is there a specific reason that the initrd partition is so large? Apr 09 04:51:05 jeanfabrice-zzz, ping Apr 09 04:51:59 ka6sox-away: pong Apr 09 06:03:22 evening Apr 09 06:04:14 ~slugtime Apr 09 06:04:16 methinks slugtime is 40 hour days 10 hours in 4 timezones with overlap Apr 09 06:04:22 hi there. Apr 09 06:10:50 anyone using the nslu2 together with a wrt54gs ? Apr 09 06:14:26 is Josh Parsons = jp30 ? Apr 09 06:20:01 03jeanfabrice * 10unslung/sources/thttpdphp/ (9 files): First realease of thttpd php package. To be tested Apr 09 06:20:40 03jeanfabrice * 10unslung/make/thttpdphp.mk: Makefile for thttpdphp Apr 09 06:22:18 03jeanfabrice * 10unslung/Makefile: Added thttpdphp in CROSS_PACKAGES_READY_FOR_TESTING Apr 09 06:23:40 jeanfabrice, yes Apr 09 06:24:50 03jeanfabrice * 10unslung/Makefile: Wrong location for thttpdphp. Corrected Apr 09 06:25:27 ka6sox: ok, Thanks. Will have some questions for him ;) Apr 09 06:30:27 jeanfabrice, cool Apr 09 06:40:46 rwhitby-nslu2: ping Apr 09 08:19:36 jeanfabrice: pong Apr 09 08:34:39 03rwhitby * 10unslung/Makefile: Added some more wl500g packages. Apr 09 09:06:19 03rwhitby * 10unslung/make/freetype.mk: Now creates it's staging dir before installing into it. Apr 09 09:07:22 03rwhitby * 10unslung/Makefile: Added thttpdphp to wl500g packages. Apr 09 11:07:35 Hi. I have a strange disk issue with ipkg - and a contribution to make to the how-to's. Apr 09 12:35:48 rh2: you can create a howto page yourself (use the naming convention described on the main howto index page. Apr 09 12:36:08 when you have the page finished, let us know and we will link it onto the main page. Apr 09 12:36:15 i know, i was just using that as an endorsement to gett some info on the other issue Apr 09 12:39:04 its bizarre; for some reason, running ipkg sometimes - not always - causes the conf partition to be marked read-only. Apr 09 12:39:12 and it does not allow remounting, i have to restart the slug. Apr 09 12:39:57 bizarre. what does the syslog say? Apr 09 12:40:18 dunno...where can i find that? Apr 09 12:41:04 you can find the output of syslog in /var/log/messages - it's the first port of call when you are debugging anything in linux. Apr 09 12:42:02 I'm a bit surprised that you were messing around with linux disk partitioning, but didn't know about syslog ... Apr 09 12:42:14 i haven't dealt with unix shit in a very long time Apr 09 12:46:10 alright, i purged out messages and i'm runningipkg again to see what it says Apr 09 12:49:15 besides that, i don't believe i can create any kind of wierdness that someone else didn't already work through Apr 09 13:04:41 ok, i recreated the error and have a log... Apr 09 13:05:00 it starts with "journal block not found"... Apr 09 13:06:00 sounds like i need to fsck that partition and rebuild the journal Apr 09 13:07:35 03rwhitby * 10unslung/Makefile: Removed wxbase from wl500g packages. Apr 09 13:23:40 dammit, even after fsck, it finds no journal block. Apr 09 13:23:42 suggestions? Apr 09 13:24:10 check /etc/fstab and see if its mounted as ext3 and rw. Apr 09 13:24:44 it is Apr 09 13:25:03 sure it's an ext3 filesystem, and not just an ext2? Apr 09 13:25:07 it starts out rw, and then is marked ro during the ipkg -force-overwrite process Apr 09 13:25:11 my question too. Apr 09 13:25:13 the only difference would be the journal Apr 09 13:25:36 recreating the journal is certainly the path to follow here - so whether its ext2 or 3 seems almost immaterial. Apr 09 13:25:51 and yes, it is set to ext3 in fstab Apr 09 13:26:00 ok, good luck then. Apr 09 13:26:01 and it shows up ext3 when i run mount before running ipkg Apr 09 13:26:03 tune2fs -j Apr 09 13:27:55 is that available on the nslu2? Apr 09 13:28:21 if you can do ext3 on a slug then ya it HAS to be. Apr 09 13:28:45 where do you think it would be? Apr 09 13:29:30 which tune2fs is your friend Apr 09 13:29:42 which tune2fs says it ain't there Apr 09 13:29:56 i do have mke2fs Apr 09 13:29:59 are you root. Apr 09 13:30:01 will that work? Apr 09 13:30:02 yes Apr 09 13:30:41 interesting. Apr 09 13:31:04 do you have another linux box available to plug it into? Apr 09 13:31:28 no Apr 09 13:32:18 see if there is the e2fsprogs ipkg available and install it if it is. Apr 09 13:33:47 ok; it'll be a minute, i have to restart it first to set it back to rw Apr 09 13:34:14 is remount "on" in fstab? Apr 09 13:34:34 remount,rw? Apr 09 13:34:45 i'll check when it comes back up Apr 09 13:34:59 okay Apr 09 13:35:09 but the disk runs fine as rw, until it hits ipkg and i use the -force-overwrite flag because of an ncurses dependancy Apr 09 13:36:52 what are you trying to load that has an ncurses dependency? Apr 09 13:37:06 nearly everything Apr 09 13:38:16 ok, how can i check if remount is "on"? there are two "1"'s by the partition listing in fstab Apr 09 13:39:07 it should say remount,rw in on of the fields. Apr 09 13:39:41 dev/sda2 /share/hdd/conf ext3 defaults 1 1 Apr 09 13:39:48 i took out the leading "/" Apr 09 13:39:55 but it is now mounted rw Apr 09 13:41:07 ok, installing e2fsprogs Apr 09 13:41:08 What does 'cat /proc/mounts' have to say about /dev/sda2? Apr 09 13:41:36 rw sync Apr 09 13:41:46 jbowler, good call. Apr 09 13:41:59 like i said, it starts out fine; it blows out when doing certain (not all) ipkg installs. Apr 09 13:42:05 syslog says the journal was not found Apr 09 13:42:10 and it remounts ro Apr 09 13:42:35 there, it did it when installing e2fsprogs Apr 09 13:42:50 Downloading http://ipkg.nslu2-linux.org/feeds/unslung/unstable/e2fsprogs_1.35-3_armeb.ipk Apr 09 13:42:52 ipkg: write: Read-only file system Apr 09 13:42:52 Could not open /usr/lib/ipkg/info/e2fsprogs.list for writing: Read-only file system Apr 09 13:42:52 Configuring e2fsprogs Apr 09 13:42:52 ipkg_conf_write_status_files: Can't open status file: /usr/lib/ipkg/status for writing: Read-only file system Apr 09 13:42:53 Could not open /usr/lib/ipkg/info/e2fsprogs.list for writing: Read-only file system Apr 09 13:42:55 pkg_write_filelist pkg=e2fsprogs returned 30 Apr 09 13:43:19 if i cat /proc/mounts, its now ro,sync Apr 09 13:43:56 What is the file type (immediately before the ro,sync)? Apr 09 13:44:16 in /proc/mounts ? Apr 09 13:44:20 Yes Apr 09 13:44:27 ext3 Apr 09 13:44:33 here's what the syslog says: Apr 09 13:44:45 <41>Apr 10 00:39:50 klogd: journal_bmap: journal block not found at offset 4108 on sd(8,2) Apr 09 13:44:58 <43>Apr 10 00:39:50 klogd: Aborting journal on device sd(8,2). Apr 09 13:44:59 <42>Apr 10 00:39:51 klogd: ext3_abort called. Apr 09 13:44:59 <42>Apr 10 00:39:51 klogd: EXT3-fs abort (device sd(8,2)): ext3_journal_start: Detected aborted journal Apr 09 13:44:59 <42>Apr 10 00:39:51 klogd: Remounting filesystem read-only Apr 09 13:45:20 df /dev/sda2? Apr 09 13:45:27 you want me to run that? Apr 09 13:45:39 Yes - to find out how full the partition is. Apr 09 13:45:50 dev/sda2 202188 10180 181566 5% /share/hdd/conf Apr 09 13:47:17 so... any ideas? Apr 09 13:47:17 It sounds like a disk error, but there would normally be a message in the syslog saying so. Apr 09 13:47:29 my suspicion is that the journal needs to be recreated Apr 09 13:47:37 at least, that seems like a safe and reasonable way to proceed Apr 09 13:47:42 except that i don't have tune2fs Apr 09 13:47:46 the question that I have is why ipkg is remounting the partition. Apr 09 13:47:59 It isn't, the kernel is doing that. Apr 09 13:48:06 i agree with jbowler Apr 09 13:48:19 The kernel (file system) tries to write a journal entry and fails, so it remounts ro. Apr 09 13:48:19 ipkg is running into a journaling error and the kernel is marking the partition ro to protect it Apr 09 13:48:53 change the filesystem type to ext2 and remount it manually Apr 09 13:49:10 then the nslu2 won't recognize it as a formatted disk Apr 09 13:49:16 (or rather remount it as ext2 and then run ipkg) Apr 09 13:49:19 boot first Apr 09 13:49:37 then remount it after it is booted up and already recognizing the disk Apr 09 13:49:45 it should then see it. Apr 09 13:50:02 Right, that should work (as a temporary fix to install e2fsprogs). Apr 09 13:50:08 exactly Apr 09 13:51:04 sounds worthwhile... Apr 09 13:53:03 then once you get the e2fsprogs installed then run tune2fs -J and rebuild the journal. Apr 09 13:53:27 and then remount it as ext3 Apr 09 13:53:48 its checking disk 1; once it mounts disk 1 rw, i'll try it... Apr 09 13:55:36 jbowler: you entered a slugbug about Feb12 Switchbox is broken...has that been fixed? Apr 09 13:56:42 its slugbug #97 Apr 09 13:57:25 dammit, it isn't changing the disk type when i try to remount... Apr 09 13:58:37 do you have a USB CF drive? Apr 09 13:59:02 yes, but lets find another way first... Apr 09 13:59:08 and actually, its at work Apr 09 13:59:23 well...we are rapidly running out of options. Apr 09 13:59:36 forced unmount fails... Apr 09 13:59:49 umount /dev/sda2 Apr 09 14:00:15 can't, busy; what do i need to kill...samba maybe Apr 09 14:00:40 yes Apr 09 14:00:53 still busy; Apr 09 14:01:18 ka6sox: bug 97, dyoung fixed it with rev 1.12 of distro/conf/switchbox.conf Apr 09 14:01:52 okay I'm going to close it out as fixed. Apr 09 14:02:03 is there a single command to disable the journaling? Apr 09 14:04:58 The very risky approach is to change the ext3 in /etc/fstab to ext2. You will lose the slug if the mount fails... Apr 09 14:05:30 well...the hd data..the slug will stil boot. Apr 09 14:05:31 The alternative is to take the e2fsprogs package and unpack it by hand on another (Linux) box. I believe it's an ar archive. Apr 09 14:05:44 I think so too. Apr 09 14:07:14 what about killing processes until the partition is no longer busy - or will that be impossible because passwd is stored there? Apr 09 14:07:15 Ah. Oh, right - the recovery is to disconnect the disk, reboot, reconnect the disk and hack /etc/fstab back to ext3... Apr 09 14:07:55 jbowler, yes that would work. Apr 09 14:08:39 what about my idea? Apr 09 14:09:16 may not work as you may kill off too many processes to keep the slug functioning. Apr 09 14:09:46 how about one of you sends me the tune2fs binary? Apr 09 14:09:52 just a suggestion... Apr 09 14:09:57 but wouldn't that work? Apr 09 14:10:38 rh2 that would work unless there is an rdepends that I don't send you. Apr 09 14:10:46 lets try it Apr 09 14:10:58 again, just a suggestion. but doesn't that seem simple? Apr 09 14:11:30 let me get to a slug that I have e2fsprogs on. Apr 09 14:11:38 thx Apr 09 14:16:49 did you find it? Apr 09 14:17:25 putting my son to bed...brb Apr 09 14:22:00 I'm building e2fsprogs, this is unslung 4.x+++ I guess but I believe it will work on older versions. Apr 09 14:25:02 I have the package and was going to put it where he could get at it. Apr 09 14:26:30 ok Apr 09 14:26:37 want to dcc it to me? Apr 09 14:27:31 jbowler, tune2fs is not in the package Apr 09 14:27:42 the one I have on the slug) Apr 09 14:28:00 so remaking it would be a good idea. Apr 09 14:30:32 It's built in misc/tune2fs. Looks like a simple C prog. Apr 09 14:31:45 is it statically compiled? Apr 09 14:32:00 tune2fs: ELF 32-bit MSB executable, ARM, version 1 (ARM), for GNU/Linux 2.4.0, dynamically linked (uses shared libs), not stripped Apr 09 14:32:09 Fresh from a bitbake/unslung build. Apr 09 14:32:42 okay you have it then...can you get it to rh2? Apr 09 14:33:36 I'm not sure - no response from /msg rh2, and 'no active DCC chat' from /msg =rh2 Apr 09 14:33:50 i'm here Apr 09 14:34:20 what happens when you try to dcc it to me? Apr 09 14:35:21 I don't know - never used dcc - I'm building it at present... Apr 09 14:35:44 oh. Apr 09 14:35:50 well...do you have a gui client? Apr 09 14:36:07 actually, why don't you just e-mail it Apr 09 14:41:05 jbowler, thanks for the help. Apr 09 14:42:18 i should have it installed in a minute... Apr 09 14:43:33 so its just tune2fs -j /dev/sda2 Apr 09 14:45:04 already has a journal; is there a force-recreate...? Apr 09 14:45:45 Yes, but: WARNING: Removing an external journal from a filesystem which was not Apr 09 14:45:46 cleanly unmounted without first replaying the external journal can result in Apr 09 14:45:46 severe data loss and filesystem corruption Apr 09 14:46:26 its not an external journal anyway Apr 09 14:46:28 I believe it should be safe once the file system is mounted read-only Apr 09 14:46:54 and tune2f -fj still gives me the "already has a journal" error Apr 09 14:48:22 wait; what if i change the error behavior to continue-operation instead of remount-ro? Apr 09 14:49:27 I saw that too, but the journal still has a problem... I can't see an option to remove or recreate the journal! Apr 09 14:49:39 yeah, didn't work Apr 09 14:49:59 03rwhitby * 10unslung/make/openvpn.mk: Changed armeb to $(TARGET_ARCH) Apr 09 14:52:26 the man page doesn't say how to fix a bad journal. Apr 09 14:52:48 what about removing a journal? Apr 09 14:52:48 google for fixing a bad ext3 journal. Apr 09 14:52:58 doesn't say that either. Apr 09 14:57:55 no good google help Apr 09 14:58:51 eh; i'm going to go watch a movie; i'll bug you guys about this tomorrow Apr 09 15:00:31 I can't find anything either - weird... Apr 09 15:02:44 someone's been busy on the slugbugs .... Apr 09 15:03:00 um.... Apr 09 15:03:42 uClibc - is there any intention to move openslug to uClibc? Apr 09 15:04:13 uClibc is a supported, but not default, target for openslug Apr 09 15:04:38 glibc is the default, we support a one line change in local.conf to build for uclibc instad Apr 09 15:04:40 instead Apr 09 15:04:53 not really a necessity unless we want to keep everything in jffs2 Apr 09 15:05:03 or need a smaller footprint. Apr 09 15:05:03 Ok, so bug 90, the uClibc stat64 problem stands. I should probably push the fix... Apr 09 15:07:03 go for it. good find. Apr 09 15:08:29 size: yes, my uClibc builds stand at 3.4MByte (rootfs) for a minimal system (I have reiser4, smbfs and nfs in that) Apr 09 15:11:31 I'm changing the status of "fixed" bugs to "resolved" instead of "closed" Apr 09 15:11:44 it makes more sense than just closing them. Apr 09 15:11:58 Ah, sorry, I got that wrong. Apr 09 15:12:08 (makes it easier to find the ones that are NOT bugs or won't fix ones) Apr 09 15:13:51 I believe the intended protocol is that a fix takes the status to 'resolve', it then gets 'verified' by the original reporter and finally 'closed' Apr 09 15:14:57 okay. Apr 09 15:15:00 thanks. Apr 09 15:15:18 I'll list them as resolved for the unverified ones. Apr 09 15:16:09 is verified also a state where it is proven to be true (as opposed to unverified) Apr 09 15:16:38 before it gets "assigned" Apr 09 15:16:50 check the "quick stats" on the front page of slugbug Apr 09 15:17:01 that's the correct order of transition Apr 09 15:17:43 okay Apr 09 15:17:44 resolved means that it has been fixed, but the fix has not been verified yet. Apr 09 15:17:52 okay Apr 09 15:18:05 closed means that no more work will be done on that bug. it can be closed fixed or closed will not fix Apr 09 15:18:24 there are a fair number that have been resolved but there is no entry that it has been verified. Apr 09 15:18:33 if you're going to change them all back, you might like to turn of the email for a while :-) Apr 09 15:19:30 anything that has been resolved for at least one full release (i.e. a later release has happened) can be assumed to be verified. if someone sees it again, they can reopen it. Apr 09 15:19:43 all but 2 of them haven't been verified (by anybody but the person who fixed it) Apr 09 15:20:52 Does the assignment work? I.e. does the assignee get an email? I'll grab 90 from [g2] since I just said I would fix it. Apr 09 15:21:04 it does Apr 09 15:21:08 yeah, assignee should get an email Apr 09 15:21:30 I think you can self-assign. Apr 09 15:21:51 I tend to be brutal on closing resolved ones if the fix has been released. Apr 09 15:22:09 after you do that I'll turn off email and then change them back to closed Apr 09 15:23:01 I can't assign to me 'cause I'm not in the drop-down list ;-) Apr 09 15:24:08 ah. you need to be added to the developer group Apr 09 15:24:41 I'll do that now Apr 09 15:24:50 done Apr 09 15:25:31 beat me Apr 09 15:30:11 disabled email from slugbug and now making changes. Apr 09 15:30:22 will enable after fixing my mess Apr 09 15:44:49 * jbowler is away: asleep Apr 09 15:47:31 nite jbowler Apr 09 15:51:12 okay moved all the bugs except #104 back into "closed" Apr 09 15:51:20 (that I moved to resolved) Apr 09 16:30:45 ~slugtime Apr 09 16:30:46 from memory, slugtime is 40 hour days 10 hours in 4 timezones with overlap Apr 09 16:31:03 morning jeanfabrice Apr 09 16:31:16 hi ka6sox Apr 09 16:31:38 ~ugt Apr 09 16:31:40 rumour has it, ugt is Universial Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. Apr 09 16:32:04 but in this case..both apply Apr 09 16:34:09 still confusing ... :) Apr 09 16:34:58 it took me about 2 months to get it all straight. Apr 09 16:35:05 but I'm still learning. Apr 09 16:36:21 and my english is really good ;) Apr 09 16:37:04 I have no problem with it....better than my french! Apr 09 16:37:24 is = is not !! Apr 09 16:47:16 morningn pTweety Apr 09 16:47:25 hiho Apr 09 16:47:53 something going on? Apr 09 16:49:19 just the usual Apr 09 16:49:25 :-) Apr 09 17:17:07 jeanfabrice: thanks for thttpdphp - that will be perfect for running the server for my new Neuston MC500 Apr 09 18:19:46 rwhitby-away: thttpdphp can be improve I think, as build method is...; humm.. particular Apr 09 21:46:31 03rwhitby 07 * r1.3199.1.1 10openembedded/packages/meta/openslug-packages.bb: Added expat and php to openslug-packages Apr 09 22:00:50 03rwhitby 07 * r1.3203 10openembedded/packages/meta/openslug-packages.bb: Removed strace from openslug-packages, cause it doesn't build Apr 09 23:33:58 hi all **** ENDING LOGGING AT Sat Apr 09 23:59:56 2005