**** BEGIN LOGGING AT Fri Jan 20 02:59:58 2006 Jan 20 05:18:33 <[g2]> NAiL pong Jan 20 05:20:53 [g2]: It had something to do with RedBoot. I can't really remember what anymore ;-) Jan 20 05:21:10 I'm back to working on the ds101 Jan 20 05:21:58 <[g2]> NAiL fis init Jan 20 05:22:03 I've tried Jan 20 05:22:04 it just hang Jan 20 05:22:06 hangs Jan 20 05:22:47 Trying to write to flash it complains about flash subsystem not initialized Jan 20 05:22:52 RAM: 0x00000000-0x20000000, [0x00022850-0x1fff1000] available Jan 20 05:22:53 FLASH: 0x00000000 - 0x00000001, 0 blocks of 0x00000000 bytes each. Jan 20 05:22:56 <[g2]> NAiL your question from earlier was about fis init, is that what your question was about ? Jan 20 05:23:07 nah Jan 20 05:23:19 I don't think so. Doesn't matter much though Jan 20 05:23:28 <[g2]> NAiL looks like the FLASH addr isn't set right Jan 20 05:23:34 <[g2]> It's looking at 0x0 Jan 20 05:23:45 <[g2]> and doesn't find anything Jan 20 05:23:49 It's a DoC device, so there's some special magic involved apparently Jan 20 05:24:07 <[g2]> yes DoC is different than FLASH Jan 20 05:24:11 <[g2]> NOR FLASH Jan 20 05:25:01 <[g2]> If I get a little time later I'll try poking around in my redboot to look for some DoC config Jan 20 05:25:17 There isn't any Jan 20 05:25:37 <[g2]> Are you looking in v 2.01/2 ? Jan 20 05:25:45 cvs checkout from today Jan 20 05:26:07 <[g2]> well you don't have IAL/CSR stuff in yours either right ? Jan 20 05:26:18 nope Jan 20 05:26:23 <[g2]> well I do Jan 20 05:26:30 hmm Jan 20 05:26:45 <[g2]> so they may (but probably aren't) different regarding the DoC Jan 20 05:28:11 the main problem right now is that I cannot (whatever I do) write to the flash Jan 20 05:28:29 <[g2]> yeah obviously, you don't have the DoC driver Jan 20 05:28:53 <[g2]> I've got the same problem with the Atheros WIFI cards Jan 20 05:29:05 <[g2]> the driver isn't working Jan 20 05:29:30 <[g2]> so the PCI config see the card, but doesn't do much Jan 20 05:29:30 This is the stock RedBoot from Synology.. It has to be able to read from flash somehow ;-) Jan 20 05:29:50 <[g2]> fis load works right ? Jan 20 05:30:46 I'll test Jan 20 05:31:34 no Jan 20 05:31:39 because that requires a name Jan 20 05:31:43 and fis list outputs nothing Jan 20 05:33:09 'cause redboot can't read the fis list from DoC ;) Jan 20 09:51:11 morning koen ;-) Jan 20 09:51:33 hey NAiL Jan 20 10:21:51 what's the spell for installing the build toolchain? ipkg install ...? Jan 20 10:22:05 openslug-native Jan 20 10:22:17 thanks! couldn't seem to find it anywhere... Jan 20 10:23:47 all i've got is slugos-native - same thing? Jan 20 10:32:34 question on FEED_URIS in local.conf are the '#' literal pound signs? Jan 20 10:33:25 it seems to have worked, but ipkg now reports two garbage characters at the front of each package name. Jan 20 11:16:11 Is there a debianslug package list somewere? Jan 20 11:16:20 [g2]: ping Jan 20 11:39:46 stein, dpkg -l | more ?! Jan 20 11:52:25 caplink811_log: I'm currently using openslug... Is there a list online? Jan 20 11:52:40 ipkg list Jan 20 11:55:08 stein, sorry, associate debianslug <-> debian Jan 20 11:57:35 caplink811_log: OK, so debianslug has a more "complete" package repository than openslug? (more packages?) Jan 20 11:57:54 Q: do I have to build openslug 3.1 myself? Or is there a downloadable binary? Jan 20 11:58:30 That is, someone told me there was a 3.1 version. Jan 20 11:58:45 I think I read somewhere that you'd have to build it yourself... Check the download page... Jan 20 11:58:55 if it's released, you can download it, otherwise you'll have to build it yourself Jan 20 11:59:03 OK. Jan 20 11:59:05 thx Jan 20 12:01:39 stein, yes, go to http://packages.debian.org/stable/ Jan 20 12:37:01 <[g2]> koen pong Jan 20 12:37:53 [g2]: I suspect I went to sleep before we had exchanged any usefull info :) Jan 20 12:38:12 <[g2]> heh Jan 20 12:41:17 my vcard should be in your inbox in about 5 minutes Jan 20 12:41:26 (or whatever time the uni mailserver needs) Jan 20 12:44:12 <[g2]> koen ok cool. I need to just check about "exporting" to the NL Jan 20 12:44:17 <[g2]> and to IT Jan 20 12:47:18 ah, the joy of customs Jan 20 12:48:41 <[g2]> it is fun Jan 20 12:49:11 * koen checks if .nl is a registered terrorist country according to the us Jan 20 12:49:38 * [g2] doesn't even know where to check :( Jan 20 12:49:59 google doesn't display it when searching fo 'axis of evil' Jan 20 12:50:07 <[g2]> heh Jan 20 12:53:12 [g2]: sending to .nl isn't that big of a deal Jan 20 12:53:36 <[g2]> ByronT I didn't think so, but I'll ask first Jan 20 12:54:02 maybe they'll bill me for VAT Jan 20 12:54:26 koen: did you get billed for the cable? Jan 20 12:54:34 ByronT: nope Jan 20 12:54:58 and neither for the h5550 and hx4700 I was sent Jan 20 12:55:17 course, I did state that what I was sending was a "gift" Jan 20 12:55:49 that usually means 'inspect and bill anyway' Jan 20 12:56:09 heh Jan 20 12:57:02 I will state that mailing via US Postal service required much less work than trying to send via FedEx Jan 20 12:58:25 in case of USPS: don't mention 'university' *anywhere* on the package Jan 20 12:58:39 otherwise the local courier will try to drop it of at the uni Jan 20 13:08:03 caplink811_log: Thanks for the link :-) Jan 20 13:08:26 np, that was for sarge, testing is sid Jan 20 13:08:54 drat, eFfeM-log is in India Jan 20 13:10:13 caplink811_log: and the debianslug apt(?)-tool defaults to sarge? Jan 20 13:11:36 stein, dunno, what is in /etc/apt/sources.list by default, myself use opendebianslug aka debonaras Jan 20 13:14:10 caplink811: opendebianslug is not aka debonaras Jan 20 13:14:19 debonaras is a project, not a distro. Jan 20 13:14:31 caplink811_log: OK. Jan 20 13:14:46 i'm realy confised about the terms Jan 20 13:14:51 debonaras could well produce a mips distribution, for example. Jan 20 13:15:00 s/confised/confused Jan 20 13:15:24 rwhitby: Hmm interesting. Got a webpage? Jan 20 13:15:31 meaning deb (be) which uses openslug-2.x as bootloader Jan 20 13:15:39 www.debonaras.org ? Jan 20 13:15:52 caplink811: yes, that is opendebianslug Jan 20 13:16:21 and just "debianslug" is the LE version of that. Jan 20 13:16:38 k Jan 20 13:16:49 however, debianslug will eventually have it's kernel and rootfs built using debian tool rather than using OE Jan 20 13:19:04 when we have debianslug images released, we'll be encouraging opendebianslug users to make a choice about whether they want to use the OE rootfs, or the Debian rootfs, and then migrate to either OpenSlug or DebianSlug. Note that anyone can always continue to use OpenDebianSlug itelf (as it is just using OpenSlug and debootstrapping to armeb debian), but the project will not be encouraging using the debianslug distribution if you're going down th Jan 20 13:19:04 e debian rootfs path. Jan 20 13:20:16 rwhitby: Will/ is debonaras "big endian"? Jan 20 13:22:53 Can you use openslug-image and install debian packages (via apt/ dpkg)? Jan 20 13:24:02 stein, that describe the be way http://www.nslu2-linux.org/wiki/DebianSlug/OpenDebianSlug Jan 20 13:25:14 caplink811_log: Ohh, thanx ;-) (I should have found that page...) Jan 20 13:25:56 stein: debianslug is LE (so it can use the existing debian arm repo) Jan 20 13:26:22 the debianslug image will boot directly into a ssh netconsole debian-installer. Jan 20 13:26:40 which you then use to install debian to the attached storage device Jan 20 13:26:55 and the last step of that installs a new kernel which loads from the external storage Jan 20 13:28:15 rwhitby: Hmm. Interesting. It's amazing how so many different distros can "emerge" around such a little gadget :-) Jan 20 13:30:46 stein: yeah, we have Unslung, OpenSlug, DebianSlug and UcSlugC. Jan 20 13:31:45 each with a very different userbase Jan 20 13:32:18 rwhitby: Yeah, btw, what are the pros with UcSlugC? Why would people choose UcSlugC? Jan 20 13:32:50 no external storage needed Jan 20 13:32:59 UcSlugC is for people who want a dedicated embedded device "turnkey" distribution, with no flash or disk attached (i.e. everything, including the application software, is in internal flash). Jan 20 13:33:06 Well, I using gentoo on my desktop. I'm just waiting for gentooslug to kick off ;-) Jan 20 13:33:30 UcSlugC is really a platform for making application-specific distributions, rather than a distribution in itself (i.e. we don't release binary images for UcSlugC). Jan 20 13:34:10 The con is that it is uClibC-based, which can make porting things to it slightly more difficult (but not overwhelmingly so) Jan 20 13:34:38 rwhitby, koen: OK, I see. I'm learning things today :-) Jan 20 13:49:36 stein: did you mention non-booting images yesterday? Jan 20 13:51:23 abrewer: Hmm, don't think so... What did you have in mind? Jan 20 13:51:54 must have you confused with somebody else, sorry. Jan 20 13:52:21 stein: i'm also kinda waiting for gentooslug (-; Jan 20 13:55:34 abrewer: Yeah, I'm really happy with gentoo on my desktop. It has been a great learning experience! I'm thinking on moving from openslug to a distro with more available packages... Jan 20 13:56:08 I wonder if gentooslug will be native compile or cross compile... I guess cross for speed... Jan 20 13:59:20 stein: dunno. we need someone to translate the german howto so language-challenged people such as myself can read about it. Jan 20 14:00:01 * koen suspects it reads 'gentoo works, but takes 2 years to build from stage1' Jan 20 14:09:14 koen: it has been well over a year since we first sent the free slug to the gentoo team :-) Jan 20 14:10:38 I have a root filesystem on a flashstick that booted with openslug-2.5. It doesn't boot on 2.7. Jan 20 14:11:15 Is there something simple I need to add? It's getting past the pivot and then it doesn't boot anything on the flashstick. Jan 20 14:11:38 beewoolie-afk: kernel module versioning problem ? Jan 20 14:11:51 No. I seem to have fixed it, inadvertently. Jan 20 14:12:05 I remove the creation of .recovery and now it boots. Jan 20 14:12:10 s/remove/removed/ Jan 20 14:12:10 beewoolie-afk meant: I removed the creation of .recovery and now it boots. Jan 20 14:12:32 yep, that's a feature ;-) Jan 20 14:12:49 No, not what I mean. I removed the line in /boot/disk that creates the recovery file. Jan 20 14:12:58 It's been the most problematic feature. Jan 20 14:13:47 the final init script should remove the .recovery file automatically ... Jan 20 14:13:48 In reading the script, I'm not sure why my change fixed the problem. I removed it because I don't want that feature. Jan 20 14:13:59 Yeah, we've been around this before. Jan 20 14:14:10 I've found that it isn't reliable because I switch to debian. Jan 20 14:14:20 Moreover, it isn't necessary with an ext3 FS. Jan 20 14:14:42 oh, .recovery is not about disk fs recovery. Jan 20 14:14:43 My users have been confounded by this more than anything else. Jan 20 14:15:13 Well, it breaks often on my systems. Honestly, reading the script it makes sense. But it breaks a lot. Jan 20 14:15:18 it's so that when you don't have serial, and your rootfs on the external disk hangs, then you're back to a bootable flash system so you can fix the external disk rootfs. Jan 20 14:15:24 I don't get it. Jan 20 14:16:05 ok, add a new init script on the external fs which does "while (1) sleep 1". Jan 20 14:16:21 in /etc/init.d? Jan 20 14:16:29 (to simulate an external rootfs which no longer boots to the command line due to some error the user made) Jan 20 14:17:11 So, where are you thinking of this? S99forcefail? Jan 20 14:17:17 cycle the power, and the .recovery file ensures you are dumped back to internal flash, so you don't just keeping pivoting into a rootfs which doesn't give you a command line Jan 20 14:17:34 (don't actually add that script, it's an example of an error condition) Jan 20 14:17:41 Yeah, but that doesn't help. the user cansimply pull the HD and recover that way. Jan 20 14:18:08 right - but the .recovery file does the same thing for people who don't have easy physical access to the device. Jan 20 14:18:23 What keeps happening to me is that I see .recovery left in the root and there isn't a reason why. Jan 20 14:18:41 the reason why is that the final init script which is supposed to delete it didn't run. Jan 20 14:18:42 The system was working, it power cycles, and then it doesn't work anymore. Jan 20 14:18:54 I don't think so. I've checked for that. Jan 20 14:18:54 (which is true for debian right now, unless you add that line) Jan 20 14:19:05 I have that script. Jan 20 14:19:14 It's something in the pivot that is breaking in this case. Jan 20 14:19:18 did you check to see if .recovery existed before power cycling? Jan 20 14:19:29 If you want, I can try it again and post the bootlog. Jan 20 14:19:38 Yeah. I have to delete it every time. Jan 20 14:19:42 if it did exist, then the script to remove it didn't work. if it didn't exist, then it will not be there on the reboot Jan 20 14:19:51 No, it isn't booting. Jan 20 14:20:05 what's happening is that it is creating .recovery and then dropping back to the flash. Jan 20 14:20:18 I delete .recovery, use turnup disk... reboot, and the same thing. Jan 20 14:20:28 When I remove the creation of .recovery, it boots to debian. Jan 20 14:20:55 ok, that's a new situation that I've never seen before. all other .recovery problems that have been reported are due to the removal script not working. Jan 20 14:21:06 right. I've know that for a long time. Jan 20 14:21:19 I got this working by accident. Jan 20 14:21:28 Let me try the 'right' way again... Jan 20 14:22:13 ... /etc/init.d/checkroot.sh removes /.recovery. Jan 20 14:23:56 ...and rcS.d/S10checkroot.sh points to the checkroot script. Jan 20 14:25:00 How is the boot script supposed to know that the system us done booting and it is ok to check for .recovery? Jan 20 14:25:59 rwhitby: Shouldn't I be able to run the boot script directly, as it is done in /linuxrc to verify that the system can pivot properly? Jan 20 14:29:36 I run /linuxrc directly and it pivots to the hard drive. NP. Jan 20 14:29:51 But the .recovery file remains. Jan 20 14:30:07 Now, I reboot to capture the log. Jan 20 14:30:36 beewoolie-afk: the check for .recovery (and decision to pivot or not) is done before the creation of /.recovery Jan 20 14:31:09 it's checking to see if the .recovery file created at the start of the previous boot (not the current boot) has been removed or not Jan 20 14:31:31 I don't see it, yet. Jan 20 14:31:42 Does turnup remove .recovery? Jan 20 14:33:02 no Jan 20 14:33:19 but turnup will complain if .recovery exists, I think Jan 20 14:33:47 I think that this may have been some sort of module problem and then I was stumbling over the .recovery thing. Jan 20 14:33:56 It looks like it's booting reliably. Jan 20 14:34:13 yes, turnup definitely checks for .recovery on the target partition, and turnup will exit with an error (before changing linuxrc) if it exists Jan 20 14:34:36 Oh, there is one other thing. it was checking the disk every time because of the clock. I wonder if that was causing trouble. Jan 20 14:34:46 I changed the FS to not check and to use ext3 on bootup. Jan 20 14:34:52 that may have been the source of the problem. Jan 20 14:35:08 After the FSCK, it was rebooting. Jan 20 14:35:33 that would prevent the .recovery file from being removed. Jan 20 14:37:03 yep, that would do it Jan 20 14:37:38 so that tells me that the disk checking script also needs a line added to remove .recovery Jan 20 14:38:00 It that normal for openslug? Jan 20 14:38:22 dunno Jan 20 14:38:48 Actually, it may be that the script that performs the reboot can remove .recovery, if it's there. That's a normal exit and it would be a vaild thing to do. Jan 20 14:38:57 yep Jan 20 14:40:20 Another way to handle it would be having a way for the flash filesystem to know that the system rebooted. if it reboots more than one without removing .recovery *then* it means that something is wrong. Jan 20 14:40:56 agreed Jan 20 14:41:06 or, perhaps a user-selectable number of reboots. Jan 20 14:41:13 Part of the turnup command. Jan 20 15:06:24 I'm finally getting good at opening these nslu2 boxes. No broken tabs. Jan 20 15:08:52 beewoolie-afk: there is a certain "knack" to it. Jan 20 15:09:31 Indeed. I wasn't going to open it, but I remembered the necessary speed hack. Jan 20 15:10:35 beewoolie-afk: you mentioned multiple users that you are supporting - do you have an application that is using the slug? Jan 20 15:10:56 Web servers. Jan 20 15:11:13 I was running my new wiki on one, but it was crashing after a while. Jan 20 15:11:34 Admittedly, it was overloaded with apache, php, and mysql. Jan 20 15:12:35 mysql on the slug? beewoolie-afk must be a glutton for punishment (or at least the slug was....) :) Jan 20 15:12:54 It ran for a while. Usually a day or two. Jan 20 15:13:21 There is something wonky going on because adding NTP to the slug I use for file serving is enough to make it crash after a while. Jan 20 15:13:25 I should give a call to a MySQL corp rep - he would be quite amused that someone has mysql runing on a nslu2 Jan 20 15:14:39 I'm thinking of using two instead of one. One for mysql and another for the web server. Jan 20 15:14:49 that should ease the memory pressure. Jan 20 15:17:07 ByronT: we could demonstrate MySQL on the slug at SCALE Jan 20 15:17:15 (the MySQL guy is giving a talk there) Jan 20 15:18:40 Yipee! for double-speed slugs Jan 20 15:23:40 the question is do we want to be listed on the MySQL "Case Studies" - probable /. effect on nslu2-linux.org? Jan 20 15:24:35 ByronT: we can do that after www.nslu2-linux.org is moved to the OSUOSL server Jan 20 15:25:06 once that happens, we can go on a major PR spree :-) Jan 20 15:25:23 :) Jan 20 15:27:32 short question: seems like ftp.debian.org is down, can someone prove this? Jan 20 15:28:36 EvilDevil: it responds to nothing. Good bet it is down or there is a DNS glitch. Jan 20 15:29:28 beewoolie-afk okay, i thougt i have a network problem over here ... Jan 20 18:35:15 <[g2]> beewoolie-afk did you tweak you MySQL config ? by default MySQL takes 50-80% of free memory Jan 20 18:43:42 [g2]: I didn't do anything special. Jan 20 18:45:13 [g2]: It ran fine, as I wrote. I've not returned to that setup because I need to build a custom kernel to get modules. Jan 20 18:46:06 is anyone using an ups? Jan 20 18:48:20 <[g2]> yes Jan 20 18:48:39 EvilDevil: Yep. Jan 20 18:49:07 are you running some sort of software (e.g. nut) to watch their status? Jan 20 18:49:45 Not on the slug. I just plugged in the UPS to the desktop and it auto-configured it as if it were a battery. Jan 20 18:49:55 <[g2]> EvilDevil I'm not although it has serial / usb output Jan 20 18:50:31 Pics is not immediately available. There are subdirs that are however. Bot-prevention is all. ;) Jan 20 18:50:46 <[g2]> being as the slug and router are the only things on it it's got a while before things crash Jan 20 18:51:17 i just set up nut (http://random.networkupstools.org/) on debianslug and was wondering if anyone else is using it Jan 20 18:51:23 Yeah, same here. Normally I'll have everything down except the slug, router, and modem. Jan 20 18:53:15 i have almost all my networking-equipment (2 switches, slug, usb-harddrive) on it... only my router and voip adapter aren't on it yet Jan 20 18:53:44 I'm sharing with the desktop and its LCD, though they're brought down when I'm not around. Jan 20 18:54:45 Besides, I've learned that the cable signal itself will go down anyway during a blackout. :( Jan 20 18:55:20 did you have the problem to find a suitable "converter"... my usv only got these "kaltgeräte"-plugs... and i needed "normal" euro plugs for the power supplies Jan 20 18:57:07 kaltgerätestecker = inlet connector for non-heating apparatus Jan 20 18:57:16 I reside in NJ, US, so don't have power supply issues. The slug does have the problem that I almost did not have enough room to plug it in on the back of this UPS. Jan 20 18:58:27 http://fuwshop.f-u-w.de/images/10902S_g.jpg <- only these kind of plugs are available on my usv Jan 20 18:59:19 Ah. There are adapters around to convert those. But that bites. Jan 20 19:00:26 I would have though it'd be the other way around, setup for the plugs you stick in the back of a PC. Jan 20 19:02:03 they were too expensive... plus shipping and so on.... i created one by myself... cut the cable of an multi-outlet power strip and connected it to an inlet connector Jan 20 19:03:37 they want 30 Euro + shipping. for a 3 outlet power strip Jan 20 19:03:52 o.O Sheesh. Jan 20 19:04:30 Did it at least have a decent power-spike rating? It should for that much. Jan 20 19:05:06 Or was just an ordinary splitter? Jan 20 19:05:47 nope... http://shop.austriaprovider.at/oscommerce/product_info.php?products_id=30 no power-spike rating, nothing at all Jan 20 19:06:55 my usv did cost 30 euro, and paying the same just for the multi-outlet power strip.... would be insane ;-) Jan 20 19:07:21 Bah. Some 'enlarge photo.' Not even a descrip. But yeah. Jan 20 19:12:22 okay.. i'll set up other things now :-) **** ENDING LOGGING AT Sat Jan 21 02:59:56 2006