Unfortunately no one can be told what fun_plug is - you have to see it for yourself.
You are not logged in.
I've noticed a weird ext3 issue that I don't understand.
I used tune2fs -j to convert all my ext2 partitions to ext3 (/dev/md0, /dev/sda4, /dev/sdb4)
Every once in a while after a reboot, I notice that /dev/md0 is mounted as ext2 rather than ext3.
Even if I explicitly try to remount it as ext3 using the "-t ext3" flag, it remains mounted as ext2.
When I umount it and run " tune2fs -j /dev/md0" again, I get the message:
tune2fs 1.41.0 (10-Jul-2008)
The filesystem already has a journal.
But then when I remount the partition, it is again (properly) ext3.
I don't understand what is happening here that seems to convert it back to ext3 yet it still has its journal.
I also have not (yet) tracked down the source of this problem.
Any thoughts?
Offline
Have you modified the raidtab2web (fstab equivalent) files, to mount partitions as ext3? Unless ext3 partitions can and will be mounted as ext2.
See first post in page:
http://dns323.kood.org/forum/viewtopic. … 25&p=2
Note that (and I don't remember wher I found out this info, but this was not in the above mentioned post), after having found and modified the raidtab2web file(s), you have to save them in the flash ram so that modifications are still ther when the DNS is rebooted:
mount -t minix /dev/mtdblock0 /sys/mtd1
cp /etc/raidtab2web /sys/mtd1/.
sync
umount /sys/mtd1
Then same thing with mtdblock1 and mtd2
Offline
Yes -- it seems like the problem was that my /etc/raidtab2web file had EXT2 in it (as of course so did the flash mirrors on /dev/mtdblock0 and /dev/mtdblock1). However, the versions on HD_a4 and HD_b4 were fine (probably because I had manually changed them before but hadn't changed the /etc ones).
However, it is unclear to me how a file like /etc/raidtab2web PREVENTS mount from mounting a share as EXT3, even when you manually run it with "-t ext3". I would have thought that a file that seemingly works at the web layer couldn't possible affect a core Linux function like mount. Did they "hack" mount to first check raidtab2web?
Offline
Yes, what you describe is strange.
For me mount is a link to busybox, which recognizes the -t flag, and , I don't find any mention of raidtab2web in the binary.
In the first place I find it a strange choice to use this "web layer" file to describe system mounts...
Offline
Yes -- I don't understand how the ability of mount/busybox to mount as an ext3 vs. ext2 is affected by raidtab2web. Especially since I tried it with both /bin/mount (which is the original firmware non busy-box version) and with /ffp/bin/mount (which is a link to Fonz's busybox and almost certainly knows nothing about raidtab2web).
I'm stumped...
Offline
OK - I'm pretty sure it has nothing to do with mount.
There is code in /usr/sbin/raidstart that at least on initial boot checks raidtab2web to see whether to mount as EXT2 (vs. EXT3).
This explains why the initial mount was as ext2. It also corresponds with the following kernel message on boot-up
user.warn kernel: EXT2-fs warning (device md0): ext2_fill_super: mounting ext3 filesystem as ext2
However, I still don't understand why I couldn't remount it manually as ext3 using "-t ext3"
Offline