**** BEGIN LOGGING AT Sun Apr 22 02:59:57 2007
Apr 22 03:25:17 <RobNC> anyone in here running Debian LE or OpenSLUG?
Apr 22 03:26:05 <pierrelux> slugos ?
Apr 22 03:26:41 <Lunar_Lamp> I have debian installed on my slug.
Apr 22 03:27:20 <RobNC> sure, slugosle
Apr 22 03:27:23 <RobNC> or debian
Apr 22 03:27:49 <RobNC> I don't have an unmodified slug to try so I am wondering if it is a problem with the hardware or a problem with linux kernel for NSLU2.
Apr 22 03:28:10 <RobNC> For slugosle, just to "ipkg update; ipkg install memtester"
Apr 22 03:28:24 <RobNC> for debian, do "apt-get install memtester"
Apr 22 03:28:58 <Lunar_Lamp> What;s your problem?
Apr 22 03:30:19 <RobNC> my problem is memtest kills the USB, due to PCI DMA error.  Apparently, "memtest" is given too much memory by the linux kernel
Apr 22 03:31:11 <joshin> RobNC: Try this:
Apr 22 03:31:12 <joshin> echo 2 > /proc/sys/vm/overcommit_memory
Apr 22 03:31:12 <joshin> echo 200 > /proc/sys/vm/overcommit_ratio
Apr 22 03:31:14 <RobNC> same thing happens in slugosle.  doing "memtester" (similar program) works file, provided you only do a test of the memory that is less than what is available
Apr 22 03:31:32 <joshin> Then run your command.
Apr 22 03:32:10 <RobNC> ok... unfortunately I have a fatslug so I have to shutdown, change apex params, then reboot again to get the extended memory detected.
Apr 22 03:32:55 <RobNC> lemme try on serial console - BRB
Apr 22 03:35:25 <RobNC> joshin: seems that the values now are overcommit_memory = 0 and ratio=50
Apr 22 03:35:32 <RobNC> (before typing those commands)
Apr 22 03:35:34 <pierrelux> RobNC: I'm under slugosbe...but your problem looks interesting
Apr 22 03:35:56 <joshin> Yeah, so things might freak out when they hit 16MB of RAM used.
Apr 22 03:35:59 <RobNC> yeah, dunno if it is related to apex 2nd stage loader incorrectly detecting memory
Apr 22 03:36:01 <joshin> It is rather anoying.
Apr 22 03:36:15 <joshin> Try it and see.  :D
Apr 22 03:36:25 <RobNC> actually it's ok, my slug has 29988MiB free
Apr 22 03:36:48 <RobNC> what do the memory and ratio do (I don't understand units)
Apr 22 03:37:27 <joshin> The first allows you to have a single process use more memory than you physically have.
Apr 22 03:37:45 <joshin> The second sets the ratio at which the kernel gets unhappy.
Apr 22 03:38:30 <joshin> You can google for all the details.  :)
Apr 22 03:38:50 <joshin> My slug has not crashed or killed processes unexpectedly since setting those values.
Apr 22 03:40:01 <RobNC> joshin: ok will try.  I actually have a 256MiB slug and if I run memtests at 250M, all is fine (if I have 251+ MB free).  If I tell it to do memtest of all avail mem, it kills the PCI bus (thus USB thus kernel crashes).  If I tell Apex to tell Linux I have 192M, then do test all mem, same thing happens.
Apr 22 03:40:24 <RobNC> I've tried this with 192, 128, 96, 64,48,32,16MB parameters for Apex
Apr 22 03:40:42 * joshin wishes he was good enough at soldering to upgrade his slugs to >32
Apr 22 03:41:23 <RobNC> Simple equation.  If I provision Apex for X memory (so long X<=256), then do a memtest of Y (Y = avail memory), then it crashes.  If I do a test with Y-1, all is fine!
Apr 22 03:41:37 <RobNC> joshin: heh I have a microscope, luckily
Apr 22 03:42:03 <RobNC> this problem is occurring when I rsync, and have a huge database
Apr 22 03:42:20 <RobNC> buffering tries to use up all RAM for caching
Apr 22 03:42:36 <joshin> Try it and see.  I run Gentoo on one slug and used to have the same problem when running rsync (for syncing portage)
Apr 22 03:43:02 <joshin> What's one more crash between friends?  :D
Apr 22 03:43:21 <RobNC> That is strange, b/c with 32MB it never happens (but then again I don't have Apex pass mem size to linux kernel)
Apr 22 03:43:26 <RobNC> Ok I'll try some tests.
Apr 22 03:43:34 <joshin> Good luck!
Apr 22 03:44:34 <RobNC> OK, BRB -gonna try now
Apr 22 03:46:40 <joshin> I'm off to break out the soldering iron for some (non-smd) stuff I can handle.  :)  
Apr 22 03:46:59 <RobNC> heh yeah not so easy w/o metcal
Apr 22 03:48:12 <joshin> :)
Apr 22 03:48:18 <RobNC> joshin: still happens...  weird
Apr 22 03:48:40 <RobNC> gonna reboot and see if I have apex params or not
Apr 22 03:49:38 <RobNC> but then again, I got it to "crash" the program w/o your parameters first.  Make that kept it from being run correctly.
Apr 22 03:52:57 <RobNC> joshin - still there - what kernel are you running?
Apr 22 03:53:15 <RobNC> I'm running 2.6.18-4-ixp4xx
Apr 22 04:21:28 <RobNC> joshin - checking manpages - thanks for the tip!!  Indeed, you are right about the overcommit_memory being improperly set from default.
Apr 22 04:36:08 <joshin> Sorry RobNC.  Was away from the computer.  I'm running 2.6.20 from around a month ago.
Apr 22 04:36:31 <joshin> slugosbe either as is or with Gentoo bootstrapped
Apr 22 04:39:36 <RobNC> joshin: ok np.  From reading pages, why set overcommit_memory >100?
Apr 22 04:40:05 <RobNC> I just set it to 80, and rsync finally didn't crash with fatslug (256MB)
Apr 22 04:40:24 <RobNC> didn't try with 200 but that seemed too dangerous, might take forever to fsck
Apr 22 04:41:03 <RobNC> debian LE is still running 2.6.18.   dunno how to get latest kernel w/o apt-get
Apr 22 04:46:15 <joshin> So that you can run a job that can use X percentage of total memory.
Apr 22 04:46:42 <joshin> With a 32MB slug, sometime when compiling libtool will use 96MB of RAM.
Apr 22 04:52:32 <RobNC> ah I see.  makes sense.  but you have swap right?
Apr 22 04:53:32 <RobNC> I thought that the 200 was percentage of ram.  according to the page I saw, it said that the total space used would be "SS + k/100" where SS = swap space, and k = value for overcommit_memory
Apr 22 04:53:32 <joshin> Of course.
Apr 22 04:53:48 <RobNC> so therefore, the k doesn't include SS
Apr 22 04:53:49 <joshin> Now that would be nutty
Apr 22 04:53:56 <joshin> if it included swap space
Apr 22 04:54:12 <RobNC> yeah that would be too unpredictavble
Apr 22 04:54:26 * joshin imagines running memtest on his swapspace
Apr 22 04:55:18 <RobNC> oops I meant:  "SS + RAM * (r/100)"
Apr 22 04:55:34 <RobNC> yeah I did memtest with swapon and then quickly realized it b/c it was crawling.
Apr 22 04:55:43 <RobNC> gotta "swapoff /dev/sdXY"  first
Apr 22 04:56:31 <joshin> Well, good luck with your experiments, I'm off to bed.  
Apr 22 04:56:32 <RobNC> gotta look up how to do sudo now.  If I do memtest as mortal, it obeys settings in /proc/etc. but if I do it as root, it ignores it!
Apr 22 04:56:45 <RobNC> ok, thanks for your help.  
Apr 22 04:56:46 <joshin> source /etc/profile?
Apr 22 04:57:01 <RobNC> um, gonna put that in /etc/rc.local
Apr 22 13:58:52 <Joykill> hey guys! i have a problem with my slug: when i run VI it crashes as soon as I leave insert mode.... anyone ever experienced this?
Apr 22 13:59:15 <Joykill> i mean VI crashes, not the slug
Apr 22 14:05:01 <Naib> I did have an issue with VIM where the screen would get flooded with the last line, some error and referencing to libc
Apr 22 14:05:08 <Naib> anything in yr syslogs?
Apr 22 14:05:33 <Snotmule> well... im kinda embarassed to say that im a noob... hardly know where to look for logs
Apr 22 14:05:44 <Naib> as ROOT
Apr 22 14:05:51 <Naib> type
Apr 22 14:05:56 <Naib> tail /var/log/messages
Apr 22 14:06:05 <Naib> this will show the last 10lines of that file
Apr 22 14:06:06 <Naib> so 
Apr 22 14:06:08 <Lunar_Lamp> /var/log is the dir where logs are usually stored Snotmule :-)
Apr 22 14:06:20 <Naib> go cause VI to crash and then quickly check the logs
Apr 22 14:06:28 <Snotmule> ok :)
Apr 22 14:08:26 <Snotmule> hmm.. can i do that from another terminal window, or does it have to be the same ?
Apr 22 14:09:24 <Snotmule> all i get in the messages tail thingie is a number of logins ive done the last hour or so
Apr 22 14:10:37 <Naib> mmm so nothing so sever to log to system log then
Apr 22 14:10:49 <Naib> what VI are you using?
Apr 22 14:10:59 <Naib> the one from busybox or the separate installed one?
Apr 22 14:11:07 <Snotmule> busybox
Apr 22 14:11:11 <Snotmule> havent tried to install another
Apr 22 14:13:55 <Snotmule> thing is all i need to do is to edit a text file every now and then... is that possible to do on a windows machine and then copying it to the slug through samba?=
Apr 22 14:14:35 <Naib> you can but you have to watch out for file permission and charage returns
Apr 22 14:14:42 <Naib> you ever thought about using NANO
Apr 22 14:14:59 <Snotmule> well, ill use anythign i can get my hands on that works ;)
Apr 22 14:15:05 <Snotmule> pikg install nano?
Apr 22 14:15:08 <Snotmule> ipkg*
Apr 22 14:15:29 <Naib> maybe... been a while since I used openslug/unslung, should already been installed tho
Apr 22 14:15:38 <Naib> what is the IRC channel for debian-slug?
Apr 22 14:16:14 <Snotmule> #debian-arm
Apr 22 14:16:30 <Snotmule> hmm... seems to be empty
Apr 22 14:16:44 <Snotmule> oh its on oftc.net
Apr 22 14:16:57 <Naib> thats not it, its on a different IRC server... I have it in my history for Xchat in linux, but not in linux atm
Apr 22 14:18:23 <Snotmule> no nano installed according to ipkg list_installed
Apr 22 14:18:33 <Snotmule> or is it part of busybox?
Apr 22 14:18:39 <Naib> don't think so
Apr 22 14:18:46 <Naib> ipkg install I guess
Apr 22 14:18:52 <Naib> good little editor
Apr 22 14:19:36 <Snotmule> do i need to do anything to get it to work? i just ipkg install'ed 
Apr 22 14:20:13 <Naib> don't have todo anything else
Apr 22 14:20:46 <Snotmule> cant do nano [file with path]....
Apr 22 14:22:01 <Snotmule> no such file or directory
Apr 22 14:35:17 <Snotmule> ok, i think i managed to solve the editing issue... any reason why a program i just installed returns this: qemu-i386: no such file or directoryh
Apr 22 14:36:14 <drif> do you have your PATH environment variable set correctly?
Apr 22 14:36:38 <drif> or did just that with nano earlier
Apr 22 14:37:08 <Snotmule> path might be the thing.... oh well, more reading up :)
Apr 22 14:39:42 <Snotmule> this is just confusing... is there an easy way to handle the path environment?
Apr 22 14:39:44 <Lunar_Lamp> Snotmule, "echo $PATH"
Apr 22 14:40:13 <drif> set | grep PATH
Apr 22 14:40:16 <drif> alternative
Apr 22 14:41:21 <Snotmule> ./sbin:/bin:/usr/bin:/usr/sbin:/
Apr 22 14:41:22 <drif> export PATH=$PATH:yourownadditionshere
Apr 22 14:41:47 <Snotmule> ok, and the path i want to add is where the qemu and nano binaries are?
Apr 22 14:41:48 <drif> although export PATH=additions:$PATH is preferred order'
Apr 22 14:42:16 <drif> which OS you're running again?
Apr 22 14:42:25 <Snotmule> unslung if thats an OS :)
Apr 22 14:42:29 <drif> with unslung export PATH=/opt/sbin:/opt/bin/:$PATH
Apr 22 14:42:36 <drif> that has worked with me
Apr 22 14:42:37 <Snotmule> ok
Apr 22 14:42:42 <Snotmule> sweet, ill try that
Apr 22 14:43:00 <Lunar_Lamp> Strange that opt* isn't already in the PATH isn't it drif?
Apr 22 14:43:09 <Lunar_Lamp> Surely that should be done at install?
Apr 22 14:43:57 <Snotmule> seems like something happened at least :)
Apr 22 14:44:08 <Snotmule> error opening terminal: unknown for nano
Apr 22 14:44:37 <drif> for example set TERM='linux'
Apr 22 14:44:54 <drif> there are other alternatives but that has been most versatile choice for me
Apr 22 14:45:01 <drif> sorry
Apr 22 14:45:03 <drif> export
Apr 22 14:45:04 <drif> not set
Apr 22 14:46:13 <drif> I urge you to create /opt/profile (if it does't exist) and add those lines there so theyre preset when you login
Apr 22 14:46:20 <drif> /opt/etc/profile
Apr 22 14:46:47 <drif> and link that file to /etc/profile (ln -s /opt/etc/profile /etc/profile)
Apr 22 14:47:08 <drif> sorry about that typo above
Apr 22 14:48:01 <Snotmule> just make a file named profile with those lines?
Apr 22 14:49:48 <drif> I mean those 'export' lines
Apr 22 14:49:55 <Snotmule> okidoki
Apr 22 14:49:59 <drif> ln -s is for creating that symlink
Apr 22 14:50:07 <Snotmule> lets see if i can get nano working then
Apr 22 14:50:10 <drif> not meant to be included in profile file
Apr 22 14:53:27 <Snotmule> ah there we go, seems nano is working
Apr 22 14:53:34 <Snotmule> progress! :)
Apr 22 15:14:13 <Snotmule> *sigh* anyone have any experience with qemu?
**** ENDING LOGGING AT Mon Apr 23 02:59:57 2007