**** BEGIN LOGGING AT Sun Mar 04 02:59:58 2007 Mar 04 03:19:37 2 Mar 04 03:31:12 03bzhou * r5735 10optware/trunk/make/py-mysql.mk: py-mysql: 1.2.2b3 -> 1.2.2 Mar 04 05:46:52 03bzhou * r5736 10optware/trunk/make/py-simplejson.mk: py-simplejson: 1.5 -> 1.6 Mar 04 05:48:24 03osas * r5737 10optware/trunk/make/asterisk14.mk: asterisk14: 1.4.0 -> 1.4.1 Mar 04 09:01:56 The monotone server might be unavailable for a few minutes while it gets migrated to a new machine. Mar 04 09:09:53 ~praise dyoung for a flawless monotone server move Mar 04 09:09:58 All hail dyoung for a flawless monotone server move! Mar 04 09:10:18 thank you thankyou. ;-) **** ENDING LOGGING AT Sun Mar 04 09:44:25 2007 **** BEGIN LOGGING AT Sun Mar 04 09:44:57 2007 Mar 04 13:43:52 hi Mar 04 14:01:42 Howdy. Mar 04 14:15:01 Weird, one of my slugs (running slugosbe) just mostly hung in the middle of a rsync backup of my real server and killed the process with an "out of memory error" despite having more than 300MB of swap available on the usb disk. Log at http://www.pastebin.ca/380986 Mar 04 14:16:11 This used to happen all the time with <2.6.20 kernels but is much more rare. Mar 04 14:18:28 Kernel version is probably .704 but might be .698. Mar 04 14:19:27 does it die if you're not running emerge at the same time? Mar 04 14:22:05 the rsync is part of emerge Mar 04 14:22:17 aha Mar 04 14:22:29 It's very odd that it's hardly using any swap Mar 04 14:23:03 Yup. Been driving me slightly batty for a while but I thought that 2.6.20 had fixed it. Mar 04 14:37:24 hm.. are you talkingabout gentoo? Mar 04 14:37:58 (sync, emerge) Mar 04 14:42:54 I am talking about having built and running emerge on OpenSlug (not Gentooslug) to sync up a portage tree (and some other stuff in a subdirectory) that is causing the error. Mar 04 14:43:17 I can replicate it either in a Gentoo chroot or just by doing the rsync. Mar 04 14:44:31 */built and am running emerge/built and am running portage Mar 04 14:48:17 (Building portage on slugos was a reaction to earlier hangs in chroots under 2.6.18 and I hadn't changed my scripts to reflect what seemed to be a 2.6.20 kernel without the problem) Mar 04 14:50:27 Since the problem is intermittent, I just posted the best logs I had. I'll see if I can get a pure rsync to behave the same. Mar 04 15:39:50 Ok, was able to get a pure rsync (same directory, no chroot/emerge or other weirdness) to fail with similar kernel issues. OS is slugosbe with rootfs on a USB hard drive. I put more data in the pastebin to hopefully make it more useful to debug. Syslog/dmesg was clear of error messages before the rsync. Mar 04 15:41:02 I've had the same error untarring large files from NFS to the disk but haven't been able to capture decent logs until now. Mar 04 15:41:43 Newest and cleanest logs: http://www.pastebin.ca/381062 Mar 04 16:29:22 good morning NAiL Mar 04 16:30:12 morning :-P Mar 04 17:25:16 03bzhou * r5738 10optware/trunk/ (5 files in 4 dirs): weechat: -> 0.2.3 with scripting disabled; added to COMMON_CROSS_PACKAGES except wl500g Mar 04 17:32:02 hey, i'm trying to reduce writes to my flash drive (in the hope that it will last longer if i do). i'm following: Mar 04 17:32:03 http://www.nslu2-linux.org/wiki/Info/FlashDriveLifespanSurvey Mar 04 17:33:04 i'm on unslung 6.8 btw Mar 04 17:33:08 the first step, i have done (added noatime for the flash drives in /etc/fstab) Mar 04 17:33:23 so my /etc/fstab looks like this: http://pastebin.com/893401 Mar 04 17:34:16 the next step says to edit /etc/init.d/checkroot.sh... Mar 04 17:34:21 but i don't seem to have this file?? Mar 04 17:34:36 it does say that the reason for editing this is a bug, which was fixed in 2.7 Mar 04 17:34:52 but version 2.7 of what?? unslung? Mar 04 17:35:06 (in which case, i guess i don't need to do anything more) Mar 04 17:36:45 also, is it a good idea to also add 'nodiratime' ? i guess that this can't hurt Mar 04 17:43:28 03bzhou * r5739 10optware/trunk/make/jabberd.mk: jabberd: 1.4.4 -> 1.6.0 Mar 04 17:45:00 iwo: regarding nodiratime, yes. Mar 04 18:03:09 iwo: you are using outdated information in the wiki, I think. Mar 04 18:04:38 if you are running unslung, see the page http://www.nslu2-linux.org/wiki/Unslung/Ext3flash Mar 04 18:05:38 * mwester is not sure if iwo is running unslung or openslug, and cannot tell from the pastebin because the pastebin server is soooooo slow he cannot view the page! Mar 04 18:06:11 * mwester reads the logs more carefully, and sees that iwo is running unslung 6.8 Mar 04 18:50:21 hey mwester, thanks for the link. Mar 04 18:50:53 i'm a bit concerned about disabling swap though (since i use enhanced-ctorrent, and i _think_ it needs quite a bit of swap space) Mar 04 18:51:43 if i make the changes 'the old fashioned way' using fstab... Mar 04 18:52:02 will they still work? Mar 04 18:52:34 ah, looks like no (cos fstab is reverted) Mar 04 18:53:17 is there any way to use the .ext3flash method, then re-enable swap? Mar 04 18:56:36 either add "swapon /dev/" to a startup script, or you can edit the /etc/rc.d/rc1 file to comment out the "swapoff" line. Mar 04 19:03:22 on the pastebin front, yeah i know what you mean on the speed problems. the service is terrible (must start using a different paste bin). Mar 04 19:03:34 it's free i suppose :\ Mar 04 19:04:38 rafb.net is free too Mar 04 19:04:42 thanks for the help mwester :) Mar 04 19:05:40 limited syntax highlighters tho :( pastebin.ca is my normal alternative Mar 04 19:06:05 interface is not as good tho Mar 04 19:06:23 mwester: i don't have /etc/rc.d/rc1 yet, does it get created after i touch .ext3flash (and reboot) Mar 04 19:07:40 if so, are the changes persisted after another reboot? Mar 04 19:12:02 /etc/rc.d/rc.1 Mar 04 19:12:41 edit with extreme caution; if you make a mistake, the machine won't boot properly, and you may have to boot to flash to fix up the problem. Mar 04 19:13:08 If you are at all uncomfortable with the vi editor, I'd suggest adding the startup script to turn swap back on instead of editing. Mar 04 19:16:14 interesting, after the 'touch', /etc/rc.d/rc.1 is edited without even a reboot? Mar 04 19:16:17 cool Mar 04 19:17:25 just to make sure then :D i just add a sinlg # at the front of this line Mar 04 19:17:41 No, it's not edited, the script has the logic to check for the existence of the file that the 'touch' creates -- if it sees the file exists, it turns off swap, and remounts with noatime enabled. :) Mar 04 19:18:09 ah Mar 04 19:18:09 i see Mar 04 19:18:13 right yes, i see the if/fi now :) Mar 04 19:47:19 is it okay to use the reboot command on the slug? does this do a proper reboot? Mar 04 19:47:32 the only reason i ask, is that i don't notice it doing any of the normal beeps Mar 04 19:48:05 another thing (out of interest), if one uses OpenSlug, does the power button still work? Mar 04 19:48:51 (to turn the device off i mean) Mar 04 19:49:02 obviously it works to turn on Mar 04 19:49:37 since there is a delay to turn off (you get a beep, then some flashing lights, then the device turns off)... Mar 04 19:50:06 surely this means that a software halt is issued (not hardware)... so there must be some kind of software to support the button and issue the command on press Mar 04 19:50:17 does that software exist in openslug? Mar 04 19:52:13 I think the power button sends a ctrl-alt-delete action Mar 04 19:52:41 to make it shut down, edit /etc/inittab and replace the "ctrl-alt-delete: " with whatever you want it to do. Mar 04 19:53:01 the "reboot" and "halt" commands are more close to a controlled crash than a proper shutdown on Linksys firmware - use "DO_Shutdown" and "DO_Reboot" instead... Mar 04 19:53:22 yeah, but those don't exist in openslug, do they? Mar 04 19:53:36 Nope, but I thought iwo was running unslung. Mar 04 19:53:44 * NAiL spent days getting that code working in the kernel ;) Mar 04 19:53:55 yes, sorry i'm asking a barage of conflicting questions :) Mar 04 19:54:15 ah, I'm talking about slugos Mar 04 19:54:19 i'm on unslung, which i chose to retain the device's original features... Mar 04 19:54:27 ok, then listen to mwester :-P Mar 04 19:54:35 reboot and halt work *properly* in slugos ;) Mar 04 19:54:37 I've never touched the original firmware or unslun. Mar 04 19:54:50 s/unslun/unslung/ Mar 04 19:54:51 NAiL meant: I've never touched the original firmware or unslung. Mar 04 19:54:51 but i ended up not using any of the original features, so since then i've beem thinking - why not just use openslug Mar 04 19:55:14 If you don't use the webinterface (or don't need it), openslug is much better ;) Mar 04 19:55:27 it's faster and leaner, and more flexible. Mar 04 19:55:32 * mwester is also thinking: why not just use openslug? :D Mar 04 19:55:35 but it also requires more knowledge of linux Mar 04 19:55:41 then i though - ah, but the one 'original feature' i would still use is the power button Mar 04 19:56:00 the powerbutton will work :) Mar 04 19:56:02 That works fine! Mar 04 19:56:07 s/though/thought Mar 04 19:56:26 Works even better, because it triggers a normal Linux shutdown. Mar 04 19:56:27 IIRC, I set it up to signal PID 1 so it does ctrl-alt-delete. Mar 04 19:56:55 * NAiL actually hacked that code. It's not the version that made it into the vanilla kernel, but nevertheless :-P Mar 04 19:57:35 is there any way I can suppress this error from /find/: - Mar 04 19:57:41 "Filesystem loop detected; `/share/flash/data' has the same device number and inode as a directory which is 3 levels higher in the filesystem hierarchy." Mar 04 19:57:59 i guess it's caused by the way unslung mounts the flash drives Mar 04 19:58:27 Use a different find command -- but it's harmless, actually. Mar 04 20:36:21 Aha, running "/sbin/sysctl -w vm.overcommit_memory=1" may prove helpful to my issue... Mar 04 20:40:34 03bzhou * r5740 10optware/trunk/ (make/jabberd.mk sources/jabberd/postinst): jabberd: removed unslung specific stuff if OPTWARE_TARGET not nslu2 Mar 04 23:59:50 I wonder what the interest level would be in me making a GentooSlug stage3 available. Mar 05 00:00:09 If it would be worth the effort... Mar 05 00:02:42 (I'm slamming my Slug to test the fix I'm hoping I found by building a stage3.) Mar 05 00:54:24 joshin: uClibc or glibc? Mar 05 00:56:09 I've been running gentoo/glibc on my slug.. since...uhm.. Sep 22, 2005 (can that be right??) Mar 05 00:56:28 that's what emerge.log says anyways :) Mar 05 00:56:48 that is right.. it was before I moved Mar 05 00:57:37 I use openslug as flash.. and still 2.6.16 kernel it seems Mar 05 01:00:18 kolla: same as you glibc with openslug/slugos as flash. I keep trying uClibc and it keeps making me crazy Mar 05 01:01:04 I'm running this irssi session on my gentoo slug now. :) Mar 05 01:01:39 03bzhou * r5741 10optware/trunk/ (Makefile make/py-curl.mk): py-curl: 7.15.5.1 -> 7.16.1; no longer broken; added py25 subpackage Mar 05 01:28:29 :) **** ENDING LOGGING AT Mon Mar 05 02:59:57 2007