**** BEGIN LOGGING AT Fri Mar 25 23:59:57 2005
Mar 26 00:26:04 <CIA-1> 03rwhitby * 10unslung/make/cron.mk: Fixed the file mode of crontab
Mar 26 00:56:18 <rwhitby-away> OK, cron is now ready for testing and deployment
Mar 26 01:25:49 <bob_tm-zzz> rwhitby-away: Thanks for taking care of it
Mar 26 01:37:22 <rwhitby-away> hey bob_tm-away
Mar 26 01:38:04 <rwhitby-away> Hope you don't mind that I moved /opt/var/cron/tabs to /opt/var/cron/crontabs (cause that's where it is on most other linux distros that I've seen)
Mar 26 01:38:27 <rwhitby-away> the changeover from stock cron to our cron, and back again, seems to work fine too
Mar 26 01:38:41 <rwhitby-away> I have our cron package install a crontab which matches the stock crontab
Mar 26 01:39:00 <rwhitby-away> Does anyone know of anything in the stock firmware which modifies the crontab?
Mar 26 01:39:33 <rwhitby-away> How about the backup stuff - does that modify the crontab?  If so, we may need to leave the stock cron running ...
Mar 26 01:39:51 <rwhitby-away> Or we could have our crontab read both the stock crontab and /opt/etc/crontab ...
Mar 26 01:50:12 <rwhitby-away> Hmm - looks like drive to drive backup puts a line in /etc/crontab ....
Mar 26 01:52:50 <CIA-1> 03rwhitby * 10unslung/sources/cron/rc.cron: No longer stops the stock cron.
Mar 26 01:53:32 <CIA-1> 03rwhitby * 10unslung/sources/cron/prerm: No longer restarts the stock cron.
Mar 26 01:54:09 <CIA-1> 03rwhitby * 10unslung/sources/cron/crontab: No longer duplicates stock crontab contents.
Mar 26 01:55:18 <rwhitby-away> ok, now the unslung cron runs alongside the stock crond (luckily they are differently named executables, so they don't both get caught by killall).
Mar 26 01:55:22 <CIA-1> 03rwhitby * 10unslung/make/cron.mk: Bumped the ipk version.
Mar 26 01:57:04 <rwhitby-away> that should be more compatible with stock behaviour now ....
Mar 26 02:09:36 <bob_tm-away> ping rwhitby-away 
Mar 26 02:13:31 <bob_tm-away> rwhitby-away: I think crontab is only modified from the web interface. As I recall the crontab formats are the same, so we could do a link from the old to the new crontab location
Mar 26 02:13:46 <bob_tm-away> ...and let the new crond run
Mar 26 02:17:45 <bob_tm-away> Still - having "our" crond read both crontabs was a good idea.
Mar 26 02:44:18 <rwhitby-away> bob_tm-away: running both crons seems to work fine
Mar 26 02:44:46 <rwhitby-away> I'm wondering if a symlink would be broken by the web interface.  I guess we can test that.  Makes it harder to ipkg remove our cron then ....
Mar 26 02:45:36 <rwhitby-away> running both together (although a bit wasteful of memory) is guaranteed to be safe
Mar 26 02:48:48 <bob_tm-away> Ok - I don't run the Linksys one - it was a blessing getting rid of its beeps and checks. I see your point wrt. removal.
Mar 26 02:49:20 <dyoung> anyone know offhand what the C call is to send a break out a serial device?
Mar 26 02:49:22 <bob_tm-away> People can always divert the LS one away
Mar 26 02:49:38 <bob_tm-away> The ioctl setting?
Mar 26 02:49:40 <bob_tm-away> nope
Mar 26 02:54:21 <rwhitby-away> bob_tm-away: yeah, people can divert the stock one.
Mar 26 02:54:55 <rwhitby-away> so that's xinetd and cron, what's needed next for drop-in package infrastructure?
Mar 26 02:55:57 <rwhitby-away> I want to modify ntpdate so that it drops in a crontab with a randomly spread activation time (i.e. take the minute in which it was installed, and use that as the hourly time to update the time).
Mar 26 02:56:33 <rwhitby-away> s/ntpdate/ntpclient/
Mar 26 02:56:34 <bob_tm-away> rwhitby-away: That is a good idea being nice to the time servers
Mar 26 02:56:43 <bob_tm-away> and to yourself - better respsonse
Mar 26 02:57:05 <bob_tm-away> 99% probably update on the hour
Mar 26 02:57:13 <rwhitby-away> yeah
Mar 26 02:57:36 <rwhitby-away> my hour is most other people's half hour, so I don't suffer from that anyway ....
Mar 26 02:58:02 <bob_tm-away> True
Mar 26 02:58:05 <dyoung> I set mine to update at a weird number.
Mar 26 02:58:17 <dyoung> like 17 or something.
Mar 26 02:58:26 <bob_tm-away> Ah - primes
Mar 26 02:58:36 <rwhitby-away> 42
Mar 26 02:59:49 <bob_tm-away> :-)
Mar 26 03:00:18 <dyoung> I think my break is someplace in the depths of termios.
Mar 26 03:00:41 <bob_tm-away> rwhitby-away: What about implementing start/stop/restart in /opt/etc/init.d ?
Mar 26 03:01:19 <bob_tm-away> dyoung: You might be right - I haven't dont it for years. Check the sources of Kermit or somthing?
Mar 26 03:01:19 <rwhitby-away> the infrastructure for that is already there - it's just all the packages which need to be modifie
Mar 26 03:01:21 <rwhitby-away> modified
Mar 26 03:01:50 <bob_tm-away> Ok - chose a skeleton
Mar 26 03:02:21 <rwhitby-away> we should create a sources/template directory and put stuff there ..
Mar 26 03:02:28 <bob_tm-away> BTW: Should we make the startup files links, so a package can be disabled without deleting the startup file?
Mar 26 03:02:59 <bob_tm-away> Run levels aren't that important a factor I guess?
Mar 26 03:03:21 <rwhitby-away> I'm ambivalent on that issue ..
Mar 26 03:03:58 <bob_tm-away> Yes, I can see there are different sides to it.
Mar 26 03:06:48 <bob_tm-away> rwhitby-away: Shall we make a main function file that can be incuded from all startup files, so they can remain minimal in size and make it easier to make globals mods to their behaviour?
Mar 26 03:07:06 <rwhitby-away> sure, go for it
Mar 26 03:07:06 <dyoung> I'm gonna have to do something ugly, this platofmr doesnt support sending of breaks over this partiuclar port.
Mar 26 03:09:10 <bob_tm-away> rwhitby-away: I was thinking of taking Gentoo's and modifying. It is GPL so it should be ok.
Mar 26 03:09:42 <rwhitby-away> cool
Mar 26 03:10:22 <CIA-1> 03rwhitby * 10unslung/sources/ntpclient/postinst: Creates a default crontab file.
Mar 26 03:11:31 <CIA-1> 03rwhitby * 10unslung/sources/ntpclient/prerm: Removes the crontab file that the postinst created.
Mar 26 03:12:38 <CIA-1> 03rwhitby * 10unslung/make/ntpclient.mk: Added the postinst and prerm, and bumped the ipk version.
Mar 26 03:16:00 <CIA-1> 03rwhitby * 10unslung/sources/ntpclient/postinst: Fixed the postinst.
Mar 26 03:17:05 <CIA-1> 03rwhitby * 10unslung/sources/ntpclient/postinst: Corrected the file modes of the created crontab file.
Mar 26 03:17:07 <rwhitby-away> Seems that our cron is particular about file modes.
Mar 26 03:19:09 <rwhitby-away> I guess that crontab line should also do a hwclock -w, but I'm too lazy to work out the syntax to do both in the one crontab line
Mar 26 03:20:02 <dyoung> (ntpdate foo; hwclock -foo) > /some/log 2>&1 ?
Mar 26 03:20:58 <rwhitby-away> I was gonna put a && in there so that the hwclock is not set if the ntpclient fails
Mar 26 03:22:19 <rwhitby-away> ah - now that cron is working, I can get logrotate working automatically ...
Mar 26 03:22:30 <rwhitby-away> and then packages can drop in logrotate files ....
Mar 26 03:24:24 <rwhitby-away> when should logrotate be run each day?
Mar 26 03:24:41 <dyoung> midnight.
Mar 26 03:24:44 <bob_tm-away> Midnight? Or 00:42 or something
Mar 26 03:24:46 <bob_tm-away> :-)
Mar 26 03:24:47 <dyoung> then logfiles are centered on the day.
Mar 26 03:25:18 <rwhitby-away> 0 0 
Mar 26 03:25:24 <bob_tm-away> I would like to make installation of syslog-ng more snappy. It is a great piece of sw.
Mar 26 03:27:17 <bob_tm-away> You should consider it.
Mar 26 03:28:05 <rwhitby-away> I used to run it on my 24x7 linux server.  I'll probably run it on the slug once it's set up.
Mar 26 03:28:17 <CIA-1> 03rwhitby * 10unslung/sources/logrotate/prerm: Now removes the crontab file.
Mar 26 03:28:41 <bob_tm-away> Tossed metalog away like a brick after finding it - finally realized cetralized logging without the noise of a proper PC
Mar 26 03:29:11 <CIA-1> 03rwhitby * 10unslung/sources/ntpclient/prerm: Fixed the rm command.
Mar 26 03:38:58 <rwhitby-away> Do we have a replacement for /bin/mail on the slug?
Mar 26 03:40:13 <CIA-1> 03rwhitby * 10unslung/ (4 files in 2 dirs): Overhauled. Generates control file. Drops in crontab file for automatic activation. Puts statefile in /opt/var/run. Bumped ipk version.
Mar 26 03:40:27 <rwhitby-away> That was logrotate
Mar 26 03:41:01 <bob_tm-away> (Y)
Mar 26 03:41:42 <CIA-1> 03rwhitby * 10unslung/sources/logrotate/postinst: Fixed the postinst.
Mar 26 03:43:57 <rwhitby-away> The default logrotate.conf rotates /var/log/messages
Mar 26 03:45:03 <rwhitby-away> Now packages can put things in /opt/etc/logrotate.d ...
Mar 26 03:45:55 <rwhitby-away> although most things should just be made to write to syslog anyway ....
Mar 26 03:47:16 <rwhitby-away> bob_tm-away, dyoung: what's outstanding for Unslung 4.x beta?
Mar 26 03:49:16 <dyoung> I'd like to get altroot kind of wedged into switchbox.
Mar 26 03:49:37 <dyoung> been lacking in time recently.
Mar 26 03:51:35 <bob_tm-away> rwhitby-away: Some device names 
Mar 26 03:53:16 <bob_tm-away> Since we do not have DEVFS_MOUNT. Can think of a few: audio, scanner, tape, all disk minor devices up 'til 10. I don't know how many are on at the moment.
Mar 26 03:53:53 <bob_tm-away> How are  we with random now?
Mar 26 03:56:28 <bob_tm-away> rwhitby-away: Heard from JP30 about pty devices?
Mar 26 03:57:36 <rwhitby-away> bob_tm-away: you worked out your bk access yet, so you can add the device names for the modules you enabled?
Mar 26 03:57:39 <bob_tm-away> rwhitby-away: Back to logging - can we force use of syslog for things like Samba?
Mar 26 03:57:57 <bob_tm-away> rwhitby-away: No - I haven't. This evening I may have time.
Mar 26 03:57:58 <rwhitby-away> built-in or replacement samba?
Mar 26 03:58:03 <bob_tm-away> Buildt in
Mar 26 03:58:25 <bob_tm-away> It would be sufficient for me to have it as a diversion script
Mar 26 03:58:29 <rwhitby-away> dunno how we're gonna change that ... we don't mess with that conf file at the moment
Mar 26 03:59:19 <rwhitby-away> I guess the default logrotate.conf should rotate the samba logs too
Mar 26 04:00:00 <bob_tm-away> Since we now have /var on disk, not ramdisk, we could go back to logging to /var rather than /opt/var ? I would still prefer not to, as then we might mix Linksys syslog output with package syslogs.
Mar 26 04:00:16 <bob_tm-away> ... if booting without disks.
Mar 26 04:02:14 <rwhitby-away> packages should put stuff in /opt/var
Mar 26 04:02:17 <bob_tm-away> rwhitby-away: Any further thoughts on how to do syslog plugin?  I'm still game for the idea of making the inittab entries scripts that can be diverted.
Mar 26 04:03:05 <bob_tm-away> As it is today, it cannot be done nicely. Just killing LS syslog and starting our own does not maintain the respawn functionality
Mar 26 04:03:30 <bob_tm-away> Killing LS syslog= removing it from inittab
Mar 26 04:03:33 <dyoung> ~seen peteru-home 
Mar 26 04:03:39 <rwhitby-away> note that /var on disk is only one of the options - there is still the option of jffs2 rootfs and packages only on disk
Mar 26 04:03:46 <jbot> peteru-home is currently on #nslu2-linux #oe
Mar 26 04:04:00 <dyoung> peteru-home, http://www.linux-wlan.org/docs/wlan_adapters.html.gz
Mar 26 04:04:23 <bob_tm-away> rwhitby-away: Yes, an even better reason to stay away from it
Mar 26 04:04:27 <dyoung> Seems like the Linksys usb G dongle is supposed to work.
Mar 26 04:07:13 <perlguru> hi
Mar 26 04:07:17 <rwhitby-away> dyoung: that's not how I read that table - the only Linksys usb dongle with linux support is B only
Mar 26 04:08:05 <bob_tm-away> rwhitby-away: Is it possible to grab hw clock time earlier than today? Would be nice to have the correct time on as many syslog entries as possible. 
Mar 26 04:09:05 <rwhitby> how many times do you really reboot a production slug though ....
Mar 26 04:09:31 <dyoung> oh yeah, youre right.  Drat.
Mar 26 04:11:30 <bob_tm-away> I was just wondering how it was done. If we made syslogd a script, we could perhaps do a diversion to fetch the time before starting e.g. syslog-ng
Mar 26 04:22:40 <rwhitby> bob_tm-away: I haven't got my head around the syslog thing.  But I'm happy to review a complete proposal on the matter if you want to write one up.
Mar 26 04:23:30 <bob_tm-away> On the list.
Mar 26 04:25:16 <rwhitby> was that a question or a statement?
Mar 26 04:25:43 <rwhitby> If the latter, I'll need a pointer ...
Mar 26 04:28:54 <bob_tm-away> It was a commitment
Mar 26 04:31:31 <rwhitby> cool, thx
Mar 26 04:32:46 <rwhitby> I'm bummed that I still haven't been able to get client certificates working.
Mar 26 04:41:27 <bob_tm-away> For what?
Mar 26 04:41:33 <bob_tm-away> thttpd?
Mar 26 04:41:45 <rwhitby> apache
Mar 26 04:41:51 <bob_tm-away> ah
Mar 26 04:42:01 <rwhitby> with an IE client
Mar 26 04:42:23 <bob_tm-away> Does it work with Firefox?
Mar 26 04:42:39 <bob_tm-away> Guess not
Mar 26 04:42:48 <rwhitby> dunno - haven't tried
Mar 26 06:13:13 <CIA-1> 03maurice 07 * r1.3182 10openembedded/ (BitKeeper/etc/logging_ok packages/meta/openslug-packages.bb): 
Mar 26 06:13:13 <CIA-1> openslug-packages.bb:
Mar 26 06:13:13 <CIA-1>  Added rsync
Mar 26 06:18:30 <rwhitby> perlguru: you might want to read up on BK_USER and BK_HOST env vars
Mar 26 06:19:21 <[g2]> morning all
Mar 26 06:19:31 <perlguru> hi [g2]
Mar 26 06:19:32 <[g2]> rwhitby, any luck with certs ?
Mar 26 06:20:23 <perlguru> ok, have to do something else. l8r!
Mar 26 06:21:20 <rwhitby> [g2]: nope
Mar 26 06:21:58 <[g2]> you're running apache 2.0.x on unslung 4.x ?
Mar 26 06:22:09 <rwhitby> yep
Mar 26 06:22:55 <[g2]> I was working on updating openslug-packages before my little hw diversion
Mar 26 06:22:56 <rwhitby> it's the old MSIE talking to OpenSSL problem I thinkk
Mar 26 07:11:37 <rwhitby> [g2]: it all works fine using "openssl s_client ..." but fails when using MSIE.
Mar 26 07:17:59 <[g2]> no joy :(
Mar 26 07:18:50 <rwhitby> plenty of people with the same problem on the net, dating back to 2002 even, but no solutions
Mar 26 07:19:23 <rwhitby> apparently it used to work (and it used to work for me with my old linux server), but then openssl changed and it hasn't worked since.
Mar 26 07:19:44 <rwhitby> (although I doubt it is openssl to blame)
Mar 26 07:20:20 <rwhitby> [Sun Mar 27 01:49:37 2005] [debug] ssl_engine_io.c(1522): OpenSSL: I/O error, 2 bytes expected to read on BIO#c6878 [mem: 215270]
Mar 26 07:20:21 <rwhitby> [Sun Mar 27 01:49:37 2005] [debug] ssl_engine_kernel.c(1809): OpenSSL: Exit: error in SSLv2 write request certificate C
Mar 26 07:20:21 <rwhitby> [Sun Mar 27 01:49:37 2005] [debug] ssl_engine_kernel.c(1809): OpenSSL: Exit: error in SSLv2 write request certificate C
Mar 26 07:20:21 <rwhitby> [Sun Mar 27 01:49:37 2005] [info] (70014)End of file found: SSL handshake interrupted by system [Hint: Stop button pressed in browser?!]
Mar 26 07:20:23 <rwhitby> [Sun Mar 27 01:49:37 2005] [info] Connection to child 1 closed with abortive shutdown(server home.rwhitby.net:443, client 192.168.1.4)
Mar 26 07:20:36 <rwhitby> that's the error signature
Mar 26 07:21:13 <rwhitby> at least tonight I've been able to prove that it works with openssl s_client - so I know I've got the server set up correctly
Mar 26 07:21:37 <rwhitby> anyway, that's me done for tonight
Mar 26 07:21:38 <rwhitby> night all
Mar 26 07:21:42 <[g2]> nite
Mar 26 08:01:18 <CIA-1> 03perlguru 07 * r1.3184 10openembedded/ (BitKeeper/etc/logging_ok packages/meta/openslug-packages.bb): 
Mar 26 08:01:18 <CIA-1> openslug-packages.bb:
Mar 26 08:01:18 <CIA-1>  Added CVS
Mar 26 09:09:24 <CIA-1> 03g2 07 * r1.3063 10openembedded/packages/meta/openslug-packages.bb: 
Mar 26 09:09:24 <CIA-1> Merge nslu2-linux@nslu2-linux.bkbits.net:openembedded
Mar 26 09:09:24 <CIA-1> into giantshoulder.com:/home/tom/dev/openslug/openembedded
Mar 26 09:09:28 <CIA-1> 03g2 07 * r1.3062 10openembedded/packages/meta/openslug-packages.bb: Update openslug packages
Mar 26 09:09:32 <CIA-1> 03g2 07OPENSLUG-0311-1 * r1.3048.1.1 10openembedded/: 
Mar 26 09:09:32 <CIA-1> Merge http://oe-devel.bkbits.net/openembedded
Mar 26 09:09:32 <CIA-1> into giantshoulder.com:/home/tom/dev/openslug/openembedded
Mar 26 12:02:45 <CIA-1> 03perlguru 07 * r1.3064.1.1 10openembedded/packages/meta/openslug-packages.bb: 
Mar 26 12:02:45 <CIA-1> openslug-packages.bb:
Mar 26 12:02:45 <CIA-1>  Added thttpd
Mar 26 18:05:16 <dwheeler> pwd
Mar 26 18:05:36 <dwheeler> do'op!
Mar 26 19:57:56 <ahmad> hi
Mar 26 19:58:45 <ahmad> i' trying to export a directory in /var using nfs, but cannot mount it
Mar 26 19:58:57 <ahmad> reason given by server: Permission denied
Mar 26 19:59:09 <ahmad> although file permissions are ok
Mar 26 19:59:12 <ahmad> any ideas??
Mar 26 19:59:34 <ahmad> can i export memory filesystem or does it have to be on disk??
Mar 26 20:02:43 <ahmad> anybody??
Mar 26 20:14:31 <ahmad> please??!!
Mar 26 20:20:00 <ahmad> log: mountd: getfh failed: Operation not permitted
Mar 26 20:20:08 <ahmad> what could be wrong??
Mar 26 20:22:01 <rwhitby-away> ahmad: no idea.  what firmware are you running for a start ...
Mar 26 20:22:10 <ahmad> unslung
Mar 26 20:22:59 <ahmad> is it possible to export a directory from memory (e.g. /var) rather than from disk??
Mar 26 20:33:24 <ahmad> BTW, samba can share it without a problem
Mar 26 20:54:00 <rwhitby-away> not sure
Mar 26 21:30:23 <peteru-home> dyoung: Thanks for the link. I ended up using a DLink DWL-G810 game adapter / bridge.
Mar 26 21:30:58 <peteru-home> It supports SuperG 108Mbps and WEP, without placing any load on the slug.
Mar 26 21:51:51 <eno> so what is the current state of python on nslu2?
Mar 26 21:59:23 <eno> the reason i'm asking is because i tried the current make/python.mk, in native, it compiles ok, but failed install
Mar 26 21:59:36 <eno> in cross it failed to build
Mar 26 21:59:57 <eno> and i found a patch at http://mail.python.org/pipermail/patches/2004-October/016312.html
Mar 26 22:00:01 <eno> which is very promising
Mar 26 22:00:07 <eno> trying it right now
Mar 26 22:49:35 <eno> just to report that the patch @ http://mail.python.org/pipermail/patches/2004-October/016312.html works smoothly, and python cross compiles
Mar 26 22:53:44 <eno> $ python2.4
Mar 26 22:53:44 <eno> Python 2.4 (#1, Mar 26 2005, 22:16:21)
Mar 26 22:53:44 <eno> [GCC 3.3.5] on linux2
Mar 26 22:53:44 <eno> Type "help", "copyright", "credits" or "license" for more information.
Mar 26 22:53:44 <eno> >>> from test import pystone
Mar 26 22:54:01 <eno> >>> pystone.main()
Mar 26 22:54:01 <eno> Pystone(1.1) time for 50000 passes = 57.1
Mar 26 22:54:01 <eno> This machine benchmarks at 875.657 pystones/second
Mar 26 22:58:49 <eno> shall i just go ahead and check in the new working python.mk?
Mar 26 23:01:15 <eno> ok, i guess i'll send those to mailing list then
Mar 26 23:02:29 <jbowler> personally I'd just check it in
Mar 26 23:04:38 <eno> alright, cvs keeps prev versions anyway
Mar 26 23:05:06 <jbowler> quite so
Mar 26 23:05:59 <jbowler> Umm...  that was incomprehensible, I meant "indeed it does"
Mar 26 23:14:03 <eno> just wish erlang or ghc cross compilation is that easy
Mar 26 23:26:46 <CIA-1> 03bzhou * 10unslung/ (3 files in 3 dirs): python ready for cross testing, no python-stage yet
**** ENDING LOGGING AT Sat Mar 26 23:59:56 2005