**** BEGIN LOGGING AT Tue Jun 21 23:59:56 2005 Jun 22 09:26:12 hi, whats the startus of bittorent clients for the nslu2? Jun 22 09:27:20 anyone here? Jun 22 09:29:12 I would search the yahoo mailing list for that information Jun 22 09:36:35 another way may be - ipkg update - ipkg list and search for an applikation with torrent in the description, If you nothing found, then the status is - no torrent client avaiable at this time Jun 22 13:59:26 bolidz, if you're looking for a vsftpd with ssl enabled, then you can it download ftom http://www.jrbnet.de/texte/vsftpd. you must have openssl installed Jun 22 13:59:43 s/ftom/from Jun 22 15:47:55 Hello... Jun 22 15:49:08 Does anyone know why "su" does not work - and only shows some output from tinylogin? Jun 22 15:50:35 I use Unslung 5.5 with some packages like SSH and Apache. Jun 22 15:54:14 Anybody here? Jun 22 21:38:41 glc_: I have made some good progress on my RAID Jun 22 21:38:44 glc_ Jun 22 21:39:33 glc: I have built the 3 arrays (data/conf/swap) but I'm seing something strange when I try to format them with mke2fs -j /mdX Jun 22 21:39:58 glc: if I do that during the reconstruction phase, it hangs the reconstruction process :( Jun 22 21:40:40 glc: and sometimes mke2fs terminates abruptly instead of finishing the format, with a "Terminated" message. Looks like it is getting a signal. Weird. Jun 22 21:43:59 glc: I'm wondering if that could be because I'm using the r25 raid kernel modules, while I'm running the r29 kernel. Jun 22 21:47:50 glc: OK I got it :) Funny. my /dev/md0 is so big that mke2fs was needing a lot of RAM to format it. Because my swap was not enabled, it was getting terminated by the OOM killer :) Jun 22 21:53:36 glc: So all I need to do is assemble my swap array first, then format it, then enable the swapon it, and then everything is OK Jun 22 22:15:51 HPReg: Sounds good. Jun 22 22:16:18 ok, i rebboted the nslu2 w/ raid for the first time Jun 22 22:16:39 Did the array survive the reboot? Jun 22 22:16:41 it mounted my raid arrays, but for some reason it seems the pivot root didn't work Jun 22 22:17:09 I'm not at the reboot step yet :) Jun 22 22:17:28 apparently the pivot root failed, so I'm still using flash as the root fs Jun 22 22:17:44 but at least all 3 arrays are started and mounted/swaponed Jun 22 22:17:58 * HPReg is happy Jun 22 22:18:11 * HPReg did not brick his NSLU2 (yet) Jun 22 22:18:51 glc: any tip on how to debug the boot process (short of playing with Set_LED and the beeper)? Jun 22 22:19:03 I need to understand why th epivot root fails Jun 22 22:19:09 You have /dev/md0 mounted on /mnt? Jun 22 22:19:21 glc: the linuxrc script does plenty of echo, where does the output go? Jun 22 22:19:39 It goes to the serial console. Jun 22 22:19:45 OK, that is it. You found it. Jun 22 22:20:25 I have mounted /dev/md0 on /share/hdd/conf too soon. I need not to mount it, becuase it cannot be mouted twice (in 2.4 kernels). That must be it. Jun 22 22:21:23 Now I understand your symlink business in your rc script Jun 22 22:21:40 I want to symlink / (the new root) to /share/hdd/data Jun 22 22:22:25 Be aware that 5.5 mounts the data partition as root, not the conf partition. Jun 22 22:22:48 glc: yeah, that is what I meant above. md0 onto data Jun 22 22:22:56 You can of course change this because you are customizing. Jun 22 22:23:45 I will be posting my diffs for rc scripts against a stock 5.5, so people can see exactly what I modified. It makes it easier to port the change to newer unslung firmware in the future Jun 22 22:24:12 I actually pivot to a nfs root. Samba serves out of /share/hdd/data with nothing actually mounted there. Jun 22 22:24:14 Anyway, back to the original topic, and for my iwn info, what is the best way to debug the boot sequence? Jun 22 22:24:36 Add a serial port. Jun 22 22:24:45 glc_: I will let samba share what is in /share/hdd/data Jun 22 22:24:52 glc: usb serial port? Jun 22 22:25:36 Yes. I use a converted nokia cell phone serial adapter blugged into a usb port on my linux machine. Jun 22 22:25:47 You can see my wiki page. Jun 22 22:26:11 glc: I think I will have to format my hd via the web interface, then copy the files it puts in the 2 partitions (data/conf) on another computer, then format the raid, then copy things back. Otherwise the web interface won't be happy and won't think I have formatted the disk Jun 22 22:26:30 glc: ok Jun 22 22:27:04 glc: and then, it /dev/console by default send to the serial port? Jun 22 22:27:10 send -> sent Jun 22 22:27:14 http://www.nslu2-linux.org/wiki/HowTo/AddASerialPort Jun 22 22:27:19 ok Jun 22 22:28:08 Yesterday, I realize with horror that syncing 2 300 GB disks to create the raid, even at 9MB/s, takes almost 10 hours :( Jun 22 22:29:21 I'm thinking of creating an extra fat32 partitions on my disks (5 GB), so I can easily share a 4.5 GB DVD with windows machines, if needed Jun 22 22:29:34 Many people have added to it, and it was merged with the pure serial conversion. Jun 22 22:29:37 4.5 -> 4.7. Dammit, i'm tired Jun 22 22:29:47 glc: ok Jun 22 22:29:56 #join gentoo Jun 22 22:30:19 Who cares whether the web interface thinks you formatted it? Jun 22 22:31:23 glc: I don't. Jun 22 22:31:51 glc: But I thought that if the files wre not there, maybe samaba would not work properly to export shares from my raid. AM i mistaken? Jun 22 22:33:04 glc: and i'm a perfectionist. I would be happy if I could completely fake 1 USB drive in Disk 1 while in fact I'm using 2 in Disk 1 and Disk 2, in raid 1 Jun 22 22:33:54 And from what I have read, the web interface only cares about files it has put on /share/hdd/[data|conf] to determine if the disk is "formatted" Jun 22 22:34:01 Samba works fine. If you disable the rc.bootbin stuff, and set the status file to have Jun 22 22:34:28 the right values it actually thinks the drives are formatted. Jun 22 22:34:54 My admin page says both drives are formatted, and I actually have no drives attached. Jun 22 22:35:12 I see. So it is just the status file. Good to know Jun 22 22:35:28 what does rc.bootbin do? (I'm looking at it now) Jun 22 22:36:11 interesting. a binary :( Jun 22 22:36:12 The status file is /etc/disk.status Jun 22 22:36:46 Run strings on it. It does a lot of things. I fake it out with ifup. Jun 22 22:38:26 interesting. What do you mean you fake it out? Jun 22 22:39:14 It mounts stuff ( bad for us ) and starts the network. Make sure you have a static IP and route setup . Jun 22 22:39:30 what can cause a flash drive not being mounted at / after /sbin/unsling and a reboot? Jun 22 22:39:30 Then you only need to run ifup. Jun 22 22:39:51 Now I understand better what this guy does. I was looking for who enables swap and so on. It is this fucking binary Jun 22 22:39:59 seems like linksys worked hard to hide things :( Jun 22 22:40:36 eno: which port is it plugged in? Jun 22 22:40:45 disk1 Jun 22 22:40:52 How big is it? Jun 22 22:40:56 1G Jun 22 22:40:57 eno: maybe the flash drive is mounted earlier in /linuxrc, so the second mount and the pivot root fails. That is what just happened to me Jun 22 22:41:15 Formatted by the web interface? Jun 22 22:41:23 right Jun 22 22:41:53 Are you sure it was mounted rw when you ran unsling? Jun 22 22:42:35 glc_: yes, unsling run successfully - no readonly warning Jun 22 22:42:54 HPReg: could you elaborate? Jun 22 22:42:55 You have /.sda1root ? Jun 22 22:43:31 eno: have you modified your /linuxrc script? If no, then what I wrote is irrelevant Jun 22 22:43:37 i do have /.sda1root Jun 22 22:43:48 what does mount show? Jun 22 22:44:28 after reboot: /dev/mtdblock4 on / type jffs2 (rw) Jun 22 22:44:33 I have a 1M sd card that unslings just fine. Jun 22 22:44:55 Does it show /dev/sda1 mounted at all? Jun 22 22:45:23 usbdevfs on /proc/bus/usb type usbdevfs (rw) Jun 22 22:45:23 /dev/sda1 on /share/hdd/data type ext3 (rw) Jun 22 22:45:23 /dev/sda2 on /share/hdd/conf type ext3 (rw,sync) Jun 22 22:45:49 yes, currently mounted on /share/hdd/data Jun 22 22:46:33 am i suppose to have /.sda1root on jffs2 root? Jun 22 22:46:55 Yes. Jun 22 22:47:13 Do you have /share/hdd/data/bin/init ? Jun 22 22:47:15 glc: your intuition was good. I just rebooted, and... # mount Jun 22 22:47:17 /dev/md0 on /initrd type jffs2 (rw) Jun 22 22:47:19 /dev/md1 on /initrd/share/hdd/conf type ext3 (rw,sync) Jun 22 22:47:21 /dev/md0 on / type ext3 (rw) Jun 22 22:47:22 yeah baby :) Jun 22 22:47:34 raid freaking rocks Jun 22 22:47:54 let me check Jun 22 22:48:07 HPReg: Welcome to the raid patrol. Jun 22 22:48:27 glc: can I have the badge :) Jun 22 22:48:31 15780 Dec 31 1969 /share/hdd/data/bin/init Jun 22 22:49:16 eno: You are running 5.5? Jun 22 22:49:29 true Jun 22 22:49:35 5.5 Jun 22 22:49:53 Do you have serial console? Jun 22 22:50:00 no i don't Jun 22 22:50:27 Is this a USB2 flash disk? Jun 22 22:51:19 yes USB2 flash drive, actually i successfully did unsling last night, i even installed a couple of packages Jun 22 22:51:35 run df Jun 22 22:51:55 how full is / ? Jun 22 22:52:02 then a friend of mine accidentally formatted it on another slug, so i'm trying to do it again Jun 22 22:52:15 rootfs 6528 5320 1208 81% / Jun 22 22:52:26 That is correct. Jun 22 22:52:46 He formatted it on a slug, or with windows? Jun 22 22:53:09 on a slug, probably under 4.20 Jun 22 22:53:52 /dev/sda1 1 818 811425 83 Linux Jun 22 22:54:12 /dev/sda3 948 1011 63488 82 Linux swap Jun 22 22:54:12 It's conf directory doesn't match your flash contents anymore then. Jun 22 22:55:10 so what do i need to do to correct that? Jun 22 22:55:41 Did you re-format it afterwards? Jun 22 22:56:14 i think i did, i can try one more time Jun 22 22:56:22 Remove your .sda1root Jun 22 22:56:24 Reboot Jun 22 22:56:29 Re-format Jun 22 22:56:32 unsling Jun 22 22:57:00 both /.sda1root and /share/hdd/data/.sda1root ? Jun 22 22:57:14 You may have to boot without the flashdisk to be able to telnet in. Jun 22 22:57:30 ok, i know how to do that Jun 22 22:58:03 /.sda1root Jun 22 22:58:26 The other one will get formatted away. Jun 22 22:58:50 right. What does .sda1root do? Jun 22 22:59:21 It tells it to pivot to sda1 as the root. Jun 22 23:01:51 formatting... Jun 22 23:03:59 done /sbin/unsling Jun 22 23:05:36 glc: i don't understand your rc.reboot: you do umount /dev/md0, but then you conitnue to exec things under /sbin and /bin. How is it possible since the root fs is gone? (I'm trying to reboot my raid cleanly) Jun 22 23:06:43 glc_: still no good Jun 22 23:07:16 HPReg: 4.0 has root in flash. /opt is stored on the drive. Jun 22 23:08:00 glc: ok. Interesting. Maybe I need to chroot to /initrd before I do that. Or pivot root again to come back to the flash... Jun 22 23:08:02 eno : It may be taking longer than 20 seconds for the flash to get mounted. Jun 22 23:09:05 Try changing cnt=20 to cnt=80 in linuxrc. Jun 22 23:09:07 it's already mounted rw at /share/hdd/data , just not / Jun 22 23:09:42 root is still jffs2: /dev/mtdblock4 on / type jffs2 (rw) Jun 22 23:10:09 HPReg. Perhaps you can remount / as ro. Jun 22 23:10:29 glc: yes. I tried on the cmd line, but it says it is busy Jun 22 23:10:30 eno: That is a different mount. Jun 22 23:10:42 maybe at rc.reboot time, it is not going to be busy Jun 22 23:11:33 HPReg After all the processes are stopped, You may be able to. mount -oremount,ro / Jun 22 23:11:42 glc: since pivot root is the normal unslung mechanism, you would hope that it is done correctly. Mmm, what is this: /usr/sbin/do_unmount in rc.local Jun 22 23:12:00 glc: yes but I should read that in the default unslung scripts, and I don't Jun 22 23:12:06 glc: hence my surprise Jun 22 23:12:19 glc: maybe nobody has ever noticed the issue, because it is ext3 Jun 22 23:12:32 so they haven't waited hours in fsck Jun 22 23:12:35 Well, the default was ramdisk. Jun 22 23:13:13 glc: the default in 5.5 is pivot root: unsling disk1 sets it up this way by creating .sda1root Jun 22 23:13:17 You didn't need to wory about ramdisk at reboot time. Jun 22 23:13:54 The default scripts came from Linksys running in ramdisk. Jun 22 23:14:22 Jffs and root disks are all enhancements. Jun 22 23:15:05 glc: would be funny if nobody has noticed the issue so far Jun 22 23:15:21 5.5 will be trickier to shutdown because you can't unmount your root. Jun 22 23:15:50 glc: yep that is why I want to remount ro. But then I cannot stop my md0 Jun 22 23:16:10 It is just a matter of shutting down processes, and halting. Jun 22 23:16:16 glc: so I think the right way probably is to pivot_root back to the flash, and then unmount and stop the raid Jun 22 23:16:54 If the md0 is ro you can just halt or reboot. Jun 22 23:17:12 glc: I guess you are right: but who knows, maybe mdadm --stop md0 writes something on the disk at the block level Jun 22 23:17:46 so it does not matter if you have remounted ro. It is not at the fs level, it is at the block level Jun 22 23:18:08 Don't attempt to stop it. Just remount ro, wait for the remount to complete, and halt. Jun 22 23:18:08 hopefully mdadm --stop is sane and it does not touch the drives at the block level Jun 22 23:18:25 glc: understood. But there is a slight risk Jun 22 23:18:55 glc: imagine that mdadm --stop writes something in its control block, somethinglike "ok, the array was stopped properly" Jun 22 23:19:11 anyway, I will try Jun 22 23:19:42 I am raided on my via box. It stops the raid just before it halts. Let me check. Jun 22 23:19:52 but gievn that unslung 5.5 uses pivot and does not remount -ro / by default, it must mean that we rely heavily on ext3 at boot time :) Jun 22 23:25:02 glc_: thanks, s/cnt=20/cnt=80/ works! Jun 22 23:27:01 eno: Thanks for the info. Flash just takes sooo long to initialize. Jun 22 23:27:59 eno: good to know Jun 22 23:28:14 My FC3 remounts everything ro before it halts. Jun 22 23:28:32 ok, it shoudl be fine then Jun 22 23:28:42 i'll do that and submit a patch :) Jun 22 23:30:16 glc : looks like the remount ro worked Jun 22 23:31:54 Good. Jun 22 23:32:57 1:35 AM here. Need to get to bed. Good night. Jun 22 23:33:11 good night. thx again for your help Jun 22 23:33:18 NP Jun 22 23:37:36 g'nite all, thanks glc for the help Jun 22 23:38:26 night eno **** ENDING LOGGING AT Wed Jun 22 23:59:57 2005