Right !

I followed the advice on here first changing the samba config file to no avail then making and mounting the swap file and running the fsck.ext3 check - again nothing doing. Finally following the good advice on here I formatted the partition /dev/mda4.

In the past the copying from an XP box timed out after bare a Gb or so … but NOW and thanks to "stilgar" and following those good instructions it WORKED!!

So far and even while I'm typing this post I've copied 30Gb and still going so suffice it to say it's going very well !

:D

So, for all you non-linux guys as long as you can follow the basic instructions listed in this post about formatting (and of course heeding the warnings) then this will most likely resolve your issue!

Adam

This doesn't look like something that can be fixed with a firmware upgrade (we don't wanted our firmware upgrades splatting our data do we :-)) so this would appear to be the only way to resolve this issue.

I feel sorry for the average Joe, no wonder WD have halved the price of these and there are so many on e-Bay.

The following commands will reformat the drive and restore the shared directories, but not their contents. All data will be lost.

 # backup the shares directory structure, # remove all files cd /shares rm -r internal/*/* tar cpsf /root/internal.tar internal # reformat the drive, don't reserve free block umount /shares/internal mkfs.ext3 -m 0 /dev/md4 mount /shares/internal # restore the shared directories tar xf /root/internal.tar rm /root/internal.tar 

Other threads mention this problem.. Maybe this info could be added to the troubleshooting page ?

http://mybookworld.wikidot.com/forum/t-50207/cannot-delete-files-read-only-dev-md4-on-mybook-we-i-1tb
http://mybookworld.wikidot.com/forum/t-52374/loosing-permissions-overnight
http://mybookworld.wikidot.com/forum/t-55527/large-file-copy-problem-solved

As I have no /var/oxsemi/smb.conf, I added the options directly to /etc/smb.conf.

- I see no difference when I turn the oplocks off.
- Adding the other options seem to make file transfers go slower.

I'm now trying http://mybookworld.wikidot.com/forum/t-55527/large-file-copy-problem-solved and hope this will work.
As I write this, alot of errors are found!

really strange because this is almost right out of the box.

I have a:
My Book World Edition I, 500GB

dmesg output:
<1>journal_bmap: journal block not found at offset 3084 on md4
<3>Aborting journal on device md4.
<2>ext3_abort called.
<2>EXT3-fs error (device md4): ext3_journal_start_sb: Detected aborted journal
<2>Remounting filesystem read-only
<4>__journal_remove_journal_head: freeing b_frozen_data

uname -a
Linux Chusuk 2.6.17.14 #1 PREEMPT Fri Jan 18 10:40:25 GMT 2008 armv5tejl GNU/Linux

I still have 5 days to bring this brick-wannabe to the shop and trade it something else, but I really like the concept of a cheapo headless linux server with a big disk :)

Anyone brave enough to try them? I've only install Samba.

Just so no-one else does the same - do NOT try booting without a swap partition….. LOL

i will put the drive in my linux box tonight and fix it…. i hope

Will

/EDIT

Ok this is wierd….. i done a scan of my network and found a MybookWorld @ 10.0.0.136 - now my mybook was renamed Earth and had a fixed ip 10.0.0.77

i managed to login with ssh as root - using my password that i had set before…..

All filesystems are back to raid and data is intact (scratches head)

Will

/EDIT #2

Its dead again LOL

/EDIT #3

and now its back again with correct hostname and IP o.O

useradd zonoskar
mkdir /shares/internal/zonoskar
chmod the directory with perms you need….

than add an entry for samba in /var/oxsemi/shares.inc

/edit: I'm at 20 GB now, so this works as far as I'm concerned. Only problem now is that I cannot connect as a user anymore, I have to use guest.

I see a large plastic brick in the making!

I copied a large file form /shares/internal to /var and got 7.3 MB/s. I guess reading/writing to the same disk is the bottleneck here. Now I'm testing the network speed and if the filesystem will be mounted as read-only again after 5GB.

/edit: Hmm, somehow I cannot connect via LAN….

i can get 12mb/s when running 'dd if=/dev/zero of=/shares/internal/big2 bs=1M count=30000' over ext3 now but that is still slow

CPU is not the issue as i can do 'dd if=/dev/zero of=/dev/sda4 bs=1M count=30000' and get 24mb/s with no problems

I am working on a way to get all raid partitions turned into ext3 in a way you suggest, i dont think it will be too hard, but we <- notice the 'we' iv'e roped you in now :D. we need to be careful that we disable any S* scripts that would try and repair the raid after a reboot.

Will

I can believe the raid is making the filesystem slower, everything probably has to be calculated by the puny 200MHz processor. Is there no way of removing the raid functions from /dev/md3 and /dev/md2 ? Maybe tar everithing in /var, recreate the partition in ext2 and restore the tar?

Yes that makes sense, did you change /etc/fstab to show the 4th partition as

/dev/sda4 instead of /dev/md4? if you leave it at 'md' it will still read/write through the raid subsystem, device mapper etc….

[root@earth ~]# more /etc/fstab

1. /etc/fstab: static file system information.

#

1. <file system> <mount pt> <type> <options> <dump> <pass>

/dev/root / ext3 rw,noauto,noatime 0 1
proc /proc proc defaults 0 0
devpts /dev/pts devpts defaults,gid=5,mode=620 0 0
sys /sys sysfs defaults 0 0
/dev/md2 none swap defaults 0 0
/dev/md3 /var ext3 defaults,noatime 0 2
/dev/sda4 /shares/internal ext3 defaults,noatime 0 2

I am determined to get write spead up to 24mb/s ……………..

Stay tuned :)

Will

I believe they installed a raid filesystem on the 1TB version to have identical behavior for the single and dual disk versions. Make sense?

If i copy my 30gb file internaly from /shares/internal to /shares/interal/WILL i see read write performance of 10mb/s in both directions (remember i changed this to e2fs)

[REDBULL@earth WILL]$ vmstat 1 30
procs ---memory-- -swap -io— —system —cpu—
r b swpd free buff cache si so bi bo in cs us sy id wa
4 3 17348 1164 356 11704 2 2 70 357 567 335 1 11 83 5
1 2 17348 972 312 12244 24 0 7448 8960 1377 610 13 63 0 24
0 2 17348 1236 304 12012 0 0 9996 10028 1349 520 2 70 0 28
0 2 17348 928 312 12316 0 0 10892 9580 1357 508 0 80 0 20
2 1 17348 820 300 12372 0 0 10700 10916 1353 491 3 71 0 26
0 1 17348 1204 304 12048 0 0 10192 11436 1357 510 1 74 0 25
2 1 17348 1168 308 12084 0 0 11148 9884 1353 477 2 81 0 17
2 1 17348 864 304 12424 0 0 10636 10612 1342 466 2 73 0 25
2 1 17348 736 304 12552 0 0 11280 9572 1360 499 2 80 0 18
2 1 17348 756 300 12540 0 0 10260 11992 1357 529 1 68 0 31
2 1 17348 732 308 12556 0 0 10256 11284 1350 505 2 77 0 21
2 2 17348 1028 308 12252 0 0 10788 11352 1347 517 0 79 0 21
0 2 17348 1176 308 12104 0 0 9612 9492 1345 507 1 72 0 27
1 2 17348 1116 312 12160 0 0 9744 9764 1325 509 0 73 0 27
2 1 17348 980 308 12300 0 0 10512 10296 1360 526 0 78 0 22
0 2 17348 1076 304 12212 0 0 11404 11596 1345 483 1 85 0 14
2 1 17348 896 304 12396 0 0 10768 9960 1364 523 0 80 0 20
2 1 17348 784 308 12504 0 0 10640 10604 1352 507 1 81 0 18
0 2 17348 848 316 12432 0 0 10896 10608 1352 491 1 81 0 18
2 1 17348 820 300 12476 0 0 10888 9428 1349 460 1 78 0 21
0 2 17348 848 300 12448 0 0 8844 10816 1308 456 1 66 0 33
2 1 17348 816 304 12472 0 0 11536 12336 1359 490 0 88 0 12
2 1 17348 972 304 12316 0 0 10384 8220 1353 483 1 78 0 21
0 2 17348 956 304 12336 0 0 9996 11784 1337 503 2 76 0 22
2 1 17348 748 312 12536 0 0 11024 9768 1351 491 3 74 0 23
0 2 17348 860 308 12368 0 0 10704 11252 1356 486 0 73 0 27
2 1 17348 932 304 12356 0 0 10320 10388 1353 502 0 78 0 22
2 1 17348 788 308 12500 0 0 10764 11276 1362 494 0 79 0 21
2 2 17348 1064 304 12228 0 0 10640 11252 1347 484 0 74 0 26
2 3 17348 1080 312 12156 0 0 11028 11516 1363 498 2 75 0 24

When i copy a file to from /shares/internal to /var (which is a raid member /dev/md3) i get 5-6mb/s

[REDBULL@earth WILL]$ vmstat 1 30
procs ---memory-- -swap -io— —system —cpu—
r b swpd free buff cache si so bi bo in cs us sy id wa
1 0 17340 1520 468 11504 2 2 105 390 571 335 1 11 82 5
0 0 17340 1512 468 11504 0 0 0 0 1172 180 3 5 93 0
2 1 17340 812 484 12092 0 0 1152 0 1109 130 0 13 82 5
2 1 17340 724 320 12308 0 0 7048 5280 1290 418 1 83 0 16
0 2 17340 956 328 12116 0 0 6284 5992 1288 438 2 73 0 25
3 2 17340 868 344 12008 0 0 6796 9632 1335 449 3 83 0 14
2 1 17340 728 340 12308 0 0 6532 3264 1271 409 1 77 0 22
2 1 17340 1136 364 11864 0 0 5392 8308 1286 467 3 70 2 25
2 2 17340 912 340 11976 0 0 6536 7164 1286 449 1 82 0 17
0 2 17340 812 344 12228 0 0 6796 3764 1282 422 3 79 0 18
2 1 17340 868 352 12160 0 0 5768 6644 1273 434 0 74 0 26
2 1 17340 728 356 12304 0 0 6920 6544 1297 415 2 86 0 12
2 1 17340 900 368 12116 0 0 6028 7004 1287 492 0 78 0 22
2 0 17340 996 340 12024 0 0 5520 6964 1288 456 0 73 0 27
2 5 17340 936 408 11720 4 0 5220 3980 1269 594 3 66 0 31
0 3 17340 1160 376 11724 24 0 5312 5656 1307 574 11 72 0 17
0 1 17340 1032 360 11844 0 0 6304 6544 1282 444 2 76 0 22
0 2 17340 836 364 12136 0 0 5772 4108 1275 483 1 67 0 32
2 1 17340 972 368 11948 28 0 4012 4668 1284 644 13 62 0 25
2 1 17340 916 336 12028 0 0 6796 5764 1287 405 0 80 0 20
2 1 17340 748 340 12224 0 0 6536 6704 1290 421 0 84 0 16
2 1 17340 892 352 11976 0 0 6556 6144 1288 420 0 83 0 17
0 1 17340 1028 360 11896 0 0 5760 6460 1290 509 1 73 0 26
2 2 17340 1080 332 11816 0 0 6668 8100 1370 448 2 83 0 15
0 1 17340 944 336 12084 0 0 6828 3724 1281 383 0 83 0 17
0 0 17340 1208 336 11828 0 0 0 0 1073 70 0 0 72 28
0 0 17340 1216 336 11828 0 0 0 0 1073 64 0 0 100 0
0 0 17340 1240 348 11828 0 0 4 3420 1130 201 0 6 85 9
0 0 17340 1240 348 11828 0 0 0 0 1073 61 0 0 100 0
0 0 17340 1240 348 11828 0 0 0 0 1072 60 0 0 100 0

Can anyone else post speeds from their MBWE, one that doesn't have these type of errors with freezes and shares suddenly going read-only

I am still pretty certain this is down to raid, i can only assume the other partitions still being raided are causing a bottleneck with I/O possibly because of samba?

Well if anyone else has any ideas i will be happy to listen

Will

I think there is RAID package from optware. Maybe you can try that.

i get 4mb/s still but this time using e2fs and not raid i get the 31gb completed, i believe there are 2 issues here,

1) the raid is broken, hence the sudden FS error and remount read only
2) poor performance through the kernel driver regarding I/O (as i know writing direct is flawless at 24mb/s)

I will work on this more this evening

I have to say im amazed that WD are not coming forward with this information, i know they would get some flak, but nowhere near as much as when i get this posted to Slashdot.org later this evening

Will

i add the PUBLIC folder and my folder 'WILL' changing permissions as needed

i restart samba

still getting 4mb a sec

oot@earth internal]# vmstat 1 30
procs ---memory-- -swap -io— —system —cpu—
r b swpd free buff cache si so bi bo in cs us sy id wa
1 1 17968 948 388 14400 55 57 133 7661 1668 803 6 76 12 6
1 1 17968 1024 352 13944 0 0 0 0 3976 1871 19 65 0 16
0 1 17968 1068 352 14312 0 0 0 4152 3462 2381 5 51 13 31
0 1 17968 1068 352 14316 0 0 0 4 3409 2585 2 41 6 51
0 1 17968 1108 348 14280 0 0 0 4100 3341 2361 3 44 8 46
1 1 17968 924 344 14476 0 0 0 4100 3547 2513 2 45 6 47
1 1 17968 1108 344 14292 0 0 0 0 3468 2586 1 44 6 50
0 1 17968 1048 332 14380 0 0 0 4112 3441 2510 2 42 11 46

I am confussed

1) i thought that if raid was slow with dd, but dd to raw disk was fast it must be poor filesystem performance?! - it would appear not, kernel perhaps?

I am about to leave work now, will look at this when i get home, does anyone have Debian running and at what speed I/O?

Regards

Will

I turned off oplock having seen your post on the WD forum :) no good im afraid

I have just fdisked /dev/sda4 and am putting e2fs on it, will mount it and see what performance i get

Thanks for your reply

(The sooner we develop our own OS for these the happier i will be. 4mb/s is terrible, my test with dd shows 24mb/s is possible)

Will

I had my samba updated with optware to 3.0.28a. Try that as last resort if you plan to return the unit.
I had successfully copied over 400G to MyBook with out it crashing. I can't get it to transfer faster then 5MB/s. I've increased TCP buffers and other samba setting. No luck.

[root@earth /]# time dd if=/dev/zero of=/shares/internal/big2 bs=1M count=30000

(after writing 8.1gb i get)

writing ‘/shares/internal/big2’: Read-only file system
7759+0 records in
7758+0 records out
8135831552 bytes (8.1 GB) copied, 735.428 seconds, 11.1 MB/s

real 12m15.880s
user 0m0.230s
sys 8m49.000s

/var/log/messages shows

Apr 27 05:13:43 earth user.alert kernel: journal_bmap: journal block not found at offset 3084 on md4
Apr 27 05:13:43 earth user.err kernel: Aborting journal on device md4.

And all that at just 11.1 MB/s

I perform the same test writing to the device directly and not through raid (DO NOT DO THIS UNLESS YOU WANT YOUR FILES TRASHED!!! i was happy to rebuild the filesystem after this operation.)

[root@earth /]# time dd if=/dev/zero of=/dev/sda4 bs=1M count=30000
30000+0 records in
30000+0 records out
31457280000 bytes (31 GB) copied, 1277.36 seconds, 24.6 MB/s

So i get avg 24.6 MB/s and it completes 31gb with no errors

I think the raid is screwed on these newer MBWE tbh

Will

i think this is raid related…. running some tests to confirm… (why they feel the need to raid a single drive is beyond me!)

Will

Well after another few gig it bombs again

Apr 27 04:29:03 earth user.alert kernel: journal_bmap: journal block not found at offset 3084 on md4
Apr 27 04:29:03 earth user.err kernel: Aborting journal on device md4.
Apr 27 04:29:03 earth user.crit kernel: ext3_abort called.
Apr 27 04:29:03 earth user.crit kernel: EXT3-fs error (device md4): ext3_journal_start_sb: Detected aborted journal
Apr 27 04:29:03 earth user.crit kernel: Remounting filesystem read-only
Apr 27 04:29:03 earth user.warn kernel: __journal_remove_journal_head: freeing b_frozen_data
Apr 27 04:29:03 earth user.warn kernel: __journal_remove_journal_head: freeing b_frozen_data

Grr

I was able to unmount and remount and continue uploading files, i am working on a perl script to perform monitoring at a deeper level to try and catch what is going wrong here

Will

I have recently bought a 1TB MBWE too and am having the same errors

I also found the western digital forum has a thread that pineapple has posted to [http://websupport.wdc.com/forum/topic.asp?TOPIC_ID=6970]

It seems as if they are no longer accepting new logins to the forum which just shows they know this is a big problem, i think this is either firmware related or OS related - i am sure the hardware is fine.

I was able to get ssh access and trash the /dev/md4 device and rebuild it and remount to /shares/internal and got 6-7gb copied over but then it died with the folloowing in /var/log/messages

Apr 26 15:14:29 earth user.alert kernel: journal_bmap: journal block not found at offset 3084 on md4
Apr 26 15:14:29 earth user.err kernel: Aborting journal on device md4.
Apr 26 15:14:29 earth user.crit kernel: ext3_abort called.
Apr 26 15:14:29 earth user.crit kernel: EXT3-fs error (device md4): ext3_journal_start_sb: Detected aborted journal
Apr 26 15:14:29 earth user.crit kernel: Remounting filesystem read-only
Apr 26 15:14:29 earth user.warn kernel: __journal_remove_journal_head: freeing b_frozen_data

I wonder what kernel people have with the 'good' MBWE ? mine is:

2.6.17.14 #1 PREEMPT Fri Jan 18 10:40:25 GMT 2008 armv5tejl GNU/Linux

I will try and get this remounted……

Will

[root@ZonoskarWorld ~]# umount /dev/md4
[root@ZonoskarWorld ~]# fsck.ext3 /dev/md4
e2fsck 1.38 (30-Jun-2005)
/dev/md4 contains a file system with errors, check forced.
Pass 1: Checking inodes, blocks, and sizes
Error allocating block bitmap (4): Memory allocation failed
e2fsck: aborted

Not enough memory?

/edit: Ok, I've added some swap (using this guide), now fsck is running (takes quite some time I think).

I bought a 1TB Mybook World yesterday. I got it home and immediately used the guides on here to setup SSH, install ushare, screen, disable cron and a load of other services to free up memory. It all worked perfectly.
I then hit a problem where my ext3 /shares/internal filesystem would freeze occasionaly and remount itself in read only mode. I ran fsck (after adding a bigger swap space) but that didn't help.
Eventually I recreated the filesystem and it's been fine ever since.
I thought I'd post this up incase anyone else was having similar problems.
The errors I was getting were : -

Apr 17 12:38:55 NAS user.alert kernel: journal_bmap: journal block not found at offset 3084 on md4
Apr 17 12:38:55 NAS user.err kernel: Aborting journal on device md4.
Apr 17 12:38:55 NAS user.crit kernel: ext3_abort called.
Apr 17 12:38:55 NAS user.crit kernel: EXT3-fs error (device md4): ext3_journal_start_sb: Detected aborted journal
Apr 17 12:38:55 NAS user.crit kernel: Remounting filesystem read-only
Apr 17 12:38:55 NAS user.warn kernel: __journal_remove_journal_head: freeing b_frozen_data

Hope this solution helps somebody else.