| 00:17.39 | *** join/#nslu2-linux rwhitby-web (www-data@netblock-66-159-209-60.dslextreme.com) |
| 00:56.54 | *** join/#nslu2-linux glc (~glc@glc.nslu2-linux) |
| 01:37.08 | *** join/#nslu2-linux [g2] (~g2@g2.nslu2-linux) |
| 04:22.33 | *** part/#nslu2-linux rwhitby-web (www-data@rwhitby.nslu2-linux) |
| 05:17.45 | *** join/#nslu2-linux Jacmet (~jacmet@50C59F0E.flatrate.dk) |
| 05:51.55 | CIA-12 | 03bzhou * 10unslung/ (make/cogito.mk sources/cogito/Makefile.patch): upstream upgrade from 0.10 to 0.11.3 |
| 05:52.18 | eno-away | nite all |
| 05:56.37 | *** join/#nslu2-linux phedny (~phedny_@dsl180-171-100.fastxdsl.nl) |
| 06:41.57 | *** join/#nslu2-linux CIA-10 (~CIA@flapjack.navi.cx) |
| 06:56.18 | *** part/#nslu2-linux AdamBaker (~aab@userch028.dsl.pipex.com) |
| 07:42.21 | *** join/#nslu2-linux senneth (senneth@senneth.staff.irssi) |
| 07:48.43 | *** join/#nslu2-linux jf-logger (~Jean-Fabr@jeanfabrice.nslu2-linux) |
| 07:55.34 | *** join/#nslu2-linux senneth (senneth@senneth.staff.irssi) |
| 08:12.18 | *** join/#nslu2-linux mr_claus (random@p549921F4.dip0.t-ipconnect.de) |
| 08:12.24 | mr_claus | morning |
| 08:12.54 | jeanfabrice | morning mr_claus |
| 08:55.11 | mr_claus | rwhitby-away: ping |
| 09:08.50 | rwhitby-away | mr_claus: whassup? |
| 10:32.55 | mr_claus | rwhitby-away: is there a way to include acl-support to the ext3 filesystem into unslung? |
| 11:30.00 | *** join/#nslu2-linux [g2] (~g2@g2.nslu2-linux) |
| 11:40.34 | *** join/#nslu2-linux caplink811 (~Joerg@dsl-082-082-081-218.arcor-ip.net) |
| 11:46.56 | *** join/#nslu2-linux Purple_Circle (~chatzilla@203-121-198-69.e-wire.net.au) |
| 11:48.05 | *** part/#nslu2-linux Purple_Circle (~chatzilla@203-121-198-69.e-wire.net.au) |
| 12:07.16 | *** join/#nslu2-linux CIA-12 (~CIA@flapjack.navi.cx) |
| 12:22.04 | rwhitby-away | mr_claus: I answered that question yesterday. Check the logs, or your IRC client backlog. The short answer is you would have to build it, test it, and prove that it was backwards compatible with stock firmware. The even shorter answer is that you probably should be running openslug if you are talking about such advanced linux features as acls. |
| 12:22.28 | rwhitby-asleep | night all |
| 12:22.56 | mr_claus | rwhitby-asleep: ok, i will check the logs, thx, i missed it yesterday |
| 12:23.29 | rwhitby-asleep | np - didn't say much more than what I just said then :-) |
| 12:41.43 | *** join/#nslu2-linux VoodooZ_Log (~VoodooZ@CPE00c0f021712c-CM014110003785.cpe.net.cable.rogers.com) |
| 13:02.28 | *** join/#nslu2-linux llagendijk (~louis@lagendijk.xs4all.nl) |
| 13:40.09 | *** join/#nslu2-linux mr_claus (random@p549921F4.dip0.t-ipconnect.de) |
| 13:52.45 | jeanfabrice | heading off home |
| 14:01.30 | *** join/#nslu2-linux Tiersten (~tman@Tiersten.nslu2-linux) |
| 15:05.08 | *** join/#nslu2-linux CIA-10 (~CIA@flapjack.navi.cx) |
| 16:25.48 | mr_claus | [g2]: do you use openslug? |
| 16:26.04 | [g2] | mr_claus, :) |
| 16:26.26 | [g2] | mr_claus, I'm the original openslug dev. |
| 16:26.38 | mr_claus | ok, good to know |
| 16:26.41 | *** join/#nslu2-linux CIA-12 (~CIA@flapjack.navi.cx) |
| 16:27.28 | mr_claus | i will try to install it and hope i can transfer my config files of unslung |
| 16:28.33 | mr_claus | does it looks like unslung, means partitioning? |
| 16:29.23 | NAiL | ka6sox: ping? |
| 16:31.52 | [g2] | mr_claus, unslung uses a different tool chain |
| 16:32.02 | [g2] | and a different kernel |
| 16:32.17 | ka6sox | [g2] IS Mr. OpenSlug |
| 16:32.26 | [g2] | heh |
| 16:32.48 | [g2] | actually jbowler is Mr. OpenSlug if you count recent changesets :) |
| 16:36.21 | ka6sox | NAiL, pong |
| 16:36.23 | NAiL | ka6sox: The mirror is still empty? |
| 16:36.36 | ka6sox | bandwidth is limited on my end. |
| 16:36.44 | NAiL | aha |
| 16:36.47 | ka6sox | I have all weekend to fill them. |
| 16:36.54 | Tiersten | Could we not do the initial rsync off SF? |
| 16:37.10 | ka6sox | Tiersten, you da man! |
| 16:37.13 | ka6sox | doh! |
| 16:37.29 | Tiersten | heh |
| 16:37.47 | NAiL | heh |
| 16:37.54 | NAiL | ok, so what's the rsync url? |
| 16:41.19 | Tiersten | erm. dunno. could always use wget I guess for the initial pull of SF |
| 16:41.37 | NAiL | ok, what's the url? :P |
| 16:42.11 | Tiersten | http://ipkg.nslu2-linux.org/ipkg/ |
| 16:43.39 | Tiersten | The complete one for everything is http://ipkg.nslu2-linux.org/feeds/ |
| 16:44.13 | Tiersten | <PROTECTED> |
| 16:45.50 | NAiL | aha... It's the complete one you'd want, isn't it? |
| 16:48.20 | ka6sox | NAiL, ya |
| 16:49.07 | NAiL | hmm.. how do I get wget to fetch all the files? |
| 16:49.20 | NAiL | wget -r only fetches the index.html file and robots.txt |
| 16:49.27 | NAiL | which is... not very useful |
| 16:52.44 | Tiersten | try again please |
| 16:53.22 | Tiersten | You'll end up getting quite a few copies of index.html but it should work |
| 16:54.30 | NAiL | hmm... why did it work now? |
| 16:54.43 | Tiersten | robots.txt was preventing it. I've temporarily renamed it |
| 16:55.03 | NAiL | aha |
| 16:55.48 | NAiL | yeah, works just like it should now |
| 16:56.23 | Tiersten | it does end up downloading loads of copies of index.html because there are links to resort the file listing but it shouldn't take very long |
| 16:57.10 | NAiL | Heh, the index.html-files are rather large. But it's finished with that part (atleast in openslug/oe) |
| 16:57.21 | ka6sox | NAiL, when your wget is done let me know so we can start others. |
| 16:57.32 | NAiL | ka6sox: will do |
| 16:57.58 | NAiL | wildly varying speeds though, from ~400kbps to ~50kbps |
| 17:00.57 | NAiL | scrap that... from ~200bps to ~450kbps |
| 17:01.12 | Tiersten | ouch |
| 17:02.48 | NAiL | <PROTECTED> |
| 17:03.55 | Tiersten | May grabbing it from SF wasn't such a hot idea |
| 17:04.21 | NAiL | There has been some problems reaching certain destinations outside Norway today... might be they're still having a bit of a problem |
| 17:05.04 | NAiL | Our primary uplink is having problems, I guess... |
| 17:05.15 | Tiersten | You're in Norway? |
| 17:05.22 | NAiL | yeah |
| 17:06.14 | NAiL | Tiersten: You're .de? Or am I completely off? ;) |
| 17:06.20 | Tiersten | .uk |
| 17:06.32 | NAiL | oh well.. not *that* far away |
| 17:06.35 | ka6sox | NAiL, I'll start another too. |
| 17:07.05 | Tiersten | Nope. Not as far as most of the others |
| 17:08.15 | NAiL | It's a bit strange though... all the index.html's and Packages files came throuh at "normal" speeds. The first ipkgs go dead slow.. |
| 17:10.18 | Tiersten | Same here |
| 17:10.24 | Tiersten | SF is being odd again it seems |
| 17:10.35 | Tiersten | It's blazing along at 281B/s |
| 17:11.07 | Tiersten | Bang goes that idea then. It's probably better to just wait for ka6sox to upload it via rsync. |
| 17:11.20 | ka6sox | nope..just let it seed from there. |
| 17:11.30 | ka6sox | much better. |
| 17:11.33 | NAiL | 19:11:19 (907.03 KB/s) - Connection closed at byte 4,096. Retrying. |
| 17:11.34 | Tiersten | ? |
| 17:11.51 | ka6sox | ugh...they must be having problems. |
| 17:12.34 | NAiL | Everything *but* the packages go fast. One or two ipkgs go fast. Then it goes dead slow... |
| 17:14.50 | ka6sox | NAiL, based on whatever is happening to Sf.net's pipe |
| 17:17.53 | NAiL | apparently. Tried from three different isps here now, same result every time |
| 17:18.07 | Tiersten | Yeah. It's SF. I get the same |
| 17:19.58 | NAiL | Will the mirrors work in a round-robin fashion, or will people just pick a mirror manually? |
| 17:20.31 | Tiersten | Sort of round robin. It'll try to distribute requests evenly between them |
| 17:20.53 | NAiL | nice |
| 17:21.01 | Tiersten | Probably will give the option to force a specific mirror as well just in case you've got some odd config or something is down |
| 17:21.39 | NAiL | good |
| 17:22.26 | mr_claus | [g2]: ok, whats happening after installing openslug firmware? telnet or ssh? |
| 17:22.39 | *** join/#nslu2-linux peteru-home (~peteru-ho@adsl-53-18.swiftdsl.com.au) |
| 17:22.42 | NAiL | ssh |
| 17:29.10 | *** join/#nslu2-linux Jacmet (~jacmet@50C59F0E.flatrate.dk) |
| 18:09.00 | *** join/#nslu2-linux caplink811_log (~caplink81@dsl-082-082-081-218.arcor-ip.net) |
| 18:22.23 | *** join/#nslu2-linux caplink811_away (~joerg@dsl-082-082-081-218.arcor-ip.net) |
| 19:08.33 | *** join/#nslu2-linux AdamBaker (~aab@userch028.dsl.pipex.com) |
| 19:39.57 | NAiL | ka6sox-*: the wget is slowly chugging along. Once in a while it fetches a lot of packages, most of the time it just waits. But it's a g77 right now :) |
| 20:02.02 | *** join/#nslu2-linux ilg (~Liviu@82.76.103.36) |
| 20:02.46 | *** part/#nslu2-linux ilg (~Liviu@82.76.103.36) |
| 20:04.23 | *** join/#nslu2-linux ka6sox-office (~tking@ka6sox.nslu2-linux) |
| 20:50.21 | *** join/#nslu2-linux caplink811_log (~caplink81@dsl-213-023-128-083.arcor-ip.net) |
| 21:16.20 | *** join/#nslu2-linux hanjo (~hanjo@pD953732D.dip.t-dialin.net) |
| 21:19.05 | jbowler-away | hanjo: I've pushed unionfs .bb files to nslu2-linux.bkbits.net |
| 21:19.29 | jbowler-away | I found it worked fine until I deleted a file, then it crashed/hung the kernel... |
| 21:19.43 | hanjo | jbowler, thx. I've read the log. |
| 21:20.14 | hanjo | jbolwer, I reverted to using the unionfs on the server side and export a single nfs root from there |
| 21:20.42 | hanjo | v .12 looks much more stable then the .10 we have in OE |
| 21:21.09 | jbowler-away | Ah, yes, server side may well be more efficient anyway. |
| 21:21.36 | hanjo | but this would not help rwhitby that wants to make an union with the jffs |
| 21:22.24 | jbowler-away | That's not particularly important - it would just be a useful thing to do. The build in the feed at present in 1.0.11 |
| 21:22.31 | hanjo | My only problem now is that somehow nfs is not triggering a proper refresh of the unionfs cache |
| 21:24.21 | hanjo | I have unions in the form dirs=/server/shared=ro:/server/slug_x=rw none /server/slug_x, with /server/slug_x nfs exported as root for slug_x |
| 21:25.28 | hanjo | but when I modify directly /server/shared on the server (i.e. install a package), the slugs are not able to see the change unless I make a remount of the union, or trigger cache purge with uniondbg |
| 21:26.49 | hanjo | once that is working, I would have a very nice diskless setup for the slugs |
| 21:27.00 | jbowler-away | One thing I didn't understand was the permissions settings - it seems that it should not be possible to write to anything except the top level directory. Maybe setting /server/shared=rw means something for the caching. |
| 21:28.15 | hanjo | I've tried several variations it did not seem to matter |
| 21:28.37 | ka6sox-office | NAiL, ping |
| 21:29.51 | hanjo | jbowler, how do you handle the /dev/ file system after turnup nfs -i ? when are they created on the server? |
| 21:31.20 | jbowler-away | hango: in earlier versions the jffs2 /dev was copied (with cpio), in the latest version turnup runs makedevs but then the first boot completes the population by adding symlinks |
| 21:32.07 | jbowler-away | If the file /dev/.permanent exists the /etc/init.d/devices script should do nothing. |
| 21:33.17 | jbowler-away | If you add /dev back into /etc/fstab as a ramfs partition then each slug will make its own. Since I think there are some named pipes in there you might have to do that. |
| 21:34.15 | jbowler-away | No, don't add it to fstab - just remove /dev/.noram and (if necessary) /dev/.permanent |
| 21:34.17 | hanjo | I am trying to modify the turnup script to customize the nfs -i step, so that only the "node specific files" like /etc/default/sysconf /etc/hostname, /var... are created on the server, since I want to fill-in the rest from the "shared" root |
| 21:34.53 | hanjo | so I was considering what to do with /dev, i.e. does it make sense to keep it on nfs or in ram |
| 21:35.21 | jbowler-away | So just run turnup -i once, from one slug and populate the shared directory, then run turnup nfs (no -i) to initialise the slugs. |
| 21:35.59 | hanjo | sure, but I need to transfer the "specific" files first. |
| 21:36.15 | jbowler-away | Most of the conffiles don't change, but you need a minimal sysconf for each slug with the MAC (to be safe), a /etc/network/interfaces and, I think, nothing else. |
| 21:36.37 | hanjo | var/lock /var/run etc can not be shared also |
| 21:36.51 | jbowler-away | If you use DHCP then the slug will fill in /etc/hostname, /etc/defaultdomain and so on. |
| 21:37.22 | hanjo | I don't have the list here (at home) but it was about 4 config files (i.e. the rsa key etc.) and the /var |
| 21:37.23 | jbowler-away | Yes - you need a tmpfs /var (i.e. you need to put that entry back into /etc/fstab). |
| 21:37.55 | jbowler-away | Ah yes - the RSA key is very important. |
| 21:38.18 | jbowler-away | It would be excessive, but you could use the code in /usr/sbin/reflash - it copies all the known configuration files. |
| 21:38.22 | hanjo | I want a per slug log, so that is why I kept the var on the server. But I might move to an tmpfs with remote logging |
| 21:39.10 | jbowler-away | You could even change /etc/syslog.conf to log to messages.hostname (but I think remote syslog is better) |
| 21:39.56 | hanjo | Yes, I use DHCP. BTW why is static still the default option in the openslug image? |
| 21:39.58 | jbowler-away | I think the code in 'reflash' does what you want, if you make it write to the new nfs root rather than the jffs2 root. |
| 21:40.16 | hanjo | I'll check it out |
| 21:40.43 | jbowler-away | DHCP: static isn't really the default, but it takes the bootproto setting from sysconf and that normally ends up as static. |
| 21:41.37 | *** join/#nslu2-linux marceln (~marceln@pion.xs4all.nl) |
| 21:41.39 | jbowler-away | If you zero the SysConf partition in flash I think you will get better results - the latest builds will pull the MAC from the last 80 bytes of the RedBoot partition. |
| 21:42.14 | hanjo | I was reconfiguring them with turnup to switch to dhcp |
| 21:42.34 | jbowler-away | So do 'ResetSysConf' on a slug, flash a relatively new build, and the result should be DHCP with minimal extra config file changes. |
| 21:43.56 | jbowler-away | Yeah, line 111 of /etc/init.d/sysconfsetup - I just write [network] and the MAC into /etc/default/sysconf |
| 21:45.20 | hanjo | Thx, I'll try that out on Monday at work. Currently I don't have any slugs at home. |
| 21:45.29 | jbowler-away | Then in /etc/default/functions the 'config' function defaults bootproto to 'dhcp', and subsequently the udhcpc script writes /etc/hostname and so on. |
| 21:46.28 | hanjo | You did a nice job with upslug, btw. |
| 21:46.34 | jbowler-away | My theory is that an out-of-the-box slug has an uninitialised SysConf, but I'm not sure about this - LinkSys may boot the darned things... |
| 21:46.55 | hanjo | No, they are initialised |
| 21:47.25 | hanjo | I have burned them with openslug straight out of the box, and it was populated |
| 21:47.37 | jbowler-away | Hum. Ok, maybe I should change this so that it always does DHCP on first boot. |
| 21:48.08 | jbowler-away | It will slow first boot a little, but 99% of people have DHCP servers these days. |
| 21:49.10 | jbowler-away | I'm really tempted not to read SysConf at all... it's just got so much bogus stuff in it. |
| 21:49.37 | hanjo | Do we need anythig more then the MAC? |
| 21:51.05 | jbowler-away | No, the other stuff is just being nice to people who have run unslung or linksys and done hostname setup etc. |
| 21:52.22 | jbowler-away | I can detect that because the IP address will have been changed from 192.168.1.77 or the disk_server_name will be set. |
| 21:52.41 | hanjo | well, that mac->hostname mapping works very well for my purposes. |
| 21:52.56 | jbowler-away | Yes, I figured it would ;-) |
| 21:53.30 | hanjo | makes scripting much easier |
| 21:53.32 | jbowler-away | With DHCP it doesn't matter because the udhcpc script overwrite /etc/hostname with the supplied value. |
| 21:54.21 | hanjo | yes |
| 21:54.36 | hanjo | Even with DHCP I will keep the same naming |
| 21:55.55 | hanjo | it is important for me to be able to walk in a room and scan the MAC barcode which gives me the hostname of the slug, and from that all the info in the inventory database |
| 21:56.20 | jbowler-away | For things like resolv.conf if you make the shared file match what gets written then udhcpc should not overwrite it - it does an md5sum of the old and new files and only writes if there is a change. |
| 21:57.53 | hanjo | but you are right, it should be up to the users to decide the naming, and giving priority to dhcp on the first boot should be a good thing |
| 21:58.24 | hanjo | reslov.conf is on shared at the moment and it works |
| 21:59.45 | hanjo | I have 10 slugs on dhcp (that mirrors the static asignment) and the rest are still on static. On 3 out of the 10 with dhcp I am playing with the unionfs stufff. |
| 22:01.18 | jbowler-away | I'm going to experiment with the DHCP stuff, at present it has a fallback to static if there is no server ('leasefail' in the udhcpc script), that should be sufficient to allow it to always run. |
| 22:02.11 | *** join/#nslu2-linux glc (~glc@glc.nslu2-linux) |
| 22:05.02 | hanjo | jbowler, that would imply that you should keep the SysConf |
| 22:06.11 | jbowler-away | Yes, but not necessarily the bootproto setting - if the IP has been changed it needs to be retained, likewise the hostname |
| 22:06.38 | hanjo | I see |
| 22:07.06 | hanjo | well, good luck. I am off to bed |
| 22:07.36 | ka6sox-office | nite hanjo |
| 22:07.52 | jbowler-away | g'nite |
| 22:07.54 | hanjo | nite all |
| 22:09.14 | hanjo | I am moving into a new apartment, and this week has been hell. Everything that can possibly hurt is hurting. I need to get some rest untill the big move tomorrow :)). |
| 22:13.30 | *** part/#nslu2-linux marceln (~marceln@pion.xs4all.nl) |
| 23:14.47 | *** join/#nslu2-linux ka6sox (~tking@adsl-66-159-198-116.dslextreme.com) |
| 23:27.21 | *** join/#nslu2-linux Tiersten (tman@Tiersten.nslu2-linux) |
| 23:39.32 | mr_claus | n8 |