**** BEGIN LOGGING AT Sun Sep 18 02:59:57 2011
Sep 18 19:00:27 <rigid> ahoy
Sep 18 19:01:01 <rigid> how do you handle "power-off" with plug-computers anyway?
Sep 18 19:01:58 <rigid> rw ubifs + sudden switchoff doesn't seem optimal to me, or is there no problem when writing no data?
Sep 18 19:02:15 <rigid> or does the NAND corrupt faster from the frequent fsck's?
Sep 18 19:02:22 <rigid> i have no clue
Sep 18 19:54:22 <eFfeM> if there is no problem with the nand the fsck will only read; this does not deteriorate flash; and if there are issues you might want to repair them anyway
Sep 18 19:54:38 <eFfeM> then again I think ubi has some provisions to avoid problems
Sep 18 19:59:22 <rigid> hmok... is there a way to avoid fsck on every boot with a plug-computer?
Sep 18 19:59:40 <rigid> i guess one way would be to use a ramfs and ro-nand access
Sep 18 20:49:32 <eFfeM> do a clean shutdown; e.g. run halt or down or shutdown or whatever it is called in your distro
Sep 18 20:53:51 <kerio> rigid: 0 0 in fstab lol
Sep 18 20:54:26 <kerio> also ext2ctl or whatever to disable fsck at fs level
Sep 18 20:56:57 <rigid> eFfeM: on a plug computer without switches? how would I run it?
Sep 18 20:57:21 <rigid> kerio: ;)
Sep 18 20:57:25 <eFfeM> e.g. remotely through ssh
Sep 18 20:57:49 <eFfeM> but it is somewhat nasty, i agree
Sep 18 20:57:54 <rigid> yep but then you need a networked pc
Sep 18 20:58:19 <rigid> best would be some kind of battery-buffer for 10s power-off or so :)
Sep 18 20:58:31 <rigid> a single "sync" call would be enough i guess
Sep 18 20:59:06 <kerio> nope
Sep 18 20:59:15 <rigid> kerio: why not?
Sep 18 20:59:16 <eFfeM> it could help to run sync periodically but if I recall correctly that is already done
Sep 18 20:59:19 <rigid> actually the capacitor in the sheeva-plug p/s buffers enough energy to do a sync-call. But how to detect the "unplugging" without hardware-hacking
Sep 18 20:59:26 <kerio> because you'd need to sigkill stuff first
Sep 18 20:59:33 <kerio> and before that, hopefully also sigterm it
Sep 18 20:59:40 <kerio> OH WAIT THAT'S WHAT INIT DOES WHEN YOU HALT
Sep 18 20:59:42 <eFfeM> if the sync occurs regularly and the system is idle probably no data will be written
Sep 18 21:00:05 <eFfeM> and otherwise put the volatile data on a sd card and replace that one if it wears out
Sep 18 21:00:15 <rigid> kerio: well, we're not talking about proper shutdown procedures...
Sep 18 21:00:57 <rigid> and no application should rely on direct i/o, so most should care about integrity und sudden death
Sep 18 21:01:01 <rigid> s/und/on
**** ENDING LOGGING AT Mon Sep 19 02:59:57 2011