Carlos E. R. wrote:

...

(4) Am I going to regret updating to 2.6.22.17-0.1 for any other reason?

You certainly should update, and not only because of this problem.

I started to run 'Online Update' in YaST, to find out what the changes are, but it's just hanging :( It puts up a window with three buttons and empty panels and an hourglass cursor and just stays like that. It doesn't refresh the window if it gets covered and doesn't go away when I close the window. Is this a known problem? 'Software Management' works OK but slowly as usual. But I'm confused by the Change Log it shows me. It starts like this: kernel-default-2.6.22.17-0.1 - The Standard Kernel for both Uniprocessor and Multiprocessor Systems 2007-09-22T13:00:00 BST - teheo@suse.de Patch name was wrong. Rename patch. - patches.drivers/libata-pata_via-kill-SATA_PATA_SHARING: Delete. - patches.drivers/libata-sata_via-kill-SATA_PATA_SHARING: sata_via: kill SATA_PATA_SHARING register handling (309069, 254158). which claims to be describing the new kernel but the first patch (which is the latest) seems to be in the time frame of the old kernel? So at the moment I don't know how to see the list of changes so I can check that this kernel does contain a fix for this problem. I also tried searching bugzilla but didn't see anything.

...

(5) I remember some recent posts about how to keep the previous kernel when updating but can't immediately find it. Can anybody point me to a how-to?

No howto that I know of. The procedure is simple, if tedious.

1) Manually download all related kernel rpms you have installed and need to upgrade - don't use yast.

You can get the installed list like this:

cer@nimrodel:~> rpm -q -a | grep kernel kernel-source-2.6.22.17-0.1 kernel-default-2.6.22.17-0.1 linux-kernel-headers-2.6.22-19 nfs-kernel-server-1.1.0-8 <== no kernel-docs-2.6.22.5-31 kernel-syms-2.6.22.17-0.1

When I do that I get: # rpm -q -a | grep kernel kernel-default-2.6.22.5-31 nfs-kernel-server-1.1.0-8

Or you can see in yast the list of what it wants to install and copy it.

2) Install manually all of them using "rpm -i list_of_kernel_rpms".

You mean to manually download the new versions of the packages that rpm -q -a listed? Is it OK to install a kernel when there's already one installed? I don't know much about this stuff :(

3) Review the /boot/grub/menu.lst to check that it is correct before rebooting.

So I need to back that up manually first. Is there anything else like that? Thanks, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Carlos E. R. wrote:

Not that I know. However... if the disk with the problem is the one used for the system... you won't be able to update.

No, the problem filesystem is an LVM volume on a separate controller as well as separate disk from the system disk. It's not even mounted, since I'm running the reiserfsck.

I think there have been several kernel updates since the one you have.

Since I can't find a list using YaST, is there a list of changes online somewhere?

Yep, you can install two kernels. It is yast who doesn't, but manually, yes, you can.

...

...

3) Review the /boot/grub/menu.lst to check that it is correct before rebooting.

...

So I need to back that up manually first. Is there anything else like that?

Not that I remember. The rpm command runs a install script that should add an entry for the new kernel, similar to the one you already have, maintaining the old one. That's what you should review, that indeed it did that, and that the files it refers to are really there, in /boot.

Thanks for that. I feel pretty confident about what to do when I do come to update the kernel now. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Howorth wrote:

I wrote:

...

I just started getting this message "No space left on device" although df indicates there is about 40 GB free on the device. It's a reiser file system on a Suse 10.3 system (the file system was originally created under 9.2 if that matters). Does anybody know what this means?

I see there was a thread in October started by Ciro Iriarte with a similar question http://lists.opensuse.org/opensuse/2007-10/msg03149.html and that had Jeff Mahoney posting a patched kernel http://lists.opensuse.org/opensuse/2007-10/msg03901.html

So that leads me to some more questions :)

(1) Am I seeing the same problem? (how do I know? - reiserfstune o/p is below if it is relevant)

(2) Did Jeff's kernel fix the problem?

(3) If so, presumably the current kernel now also fixes it? (I have 2.6.22.5-31 and YaST says 2.6.22.17-0.1 is available)

Hmm, I've now found an open bug on bugzilla https://bugzilla.novell.com/show_bug.cgi?id=353885 that might conceivably be related. So now I'm really not sure whether a kernel update will actually help. That bug report seems to be stalled. It looks like Jeff hasn't noticed that he got the data he asked for. Does anybody know anything about these reiserfs problems in 10.3? Thanks, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Carlos E. R. wrote:

The Tuesday 2008-03-11 at 16:59 -0000, Dave Howorth wrote:

...

Hmm, I've now found an open bug on bugzilla https://bugzilla.novell.com/show_bug.cgi?id=353885 that might conceivably be related. So now I'm really not sure whether a kernel update will actually help.

...

That bug report seems to be stalled. It looks like Jeff hasn't noticed that he got the data he asked for.

Somebody did a "ping" some hours ago :-)

Ah, thanks, refreshing the bug display showed me that.

Why don't you add your info to that report? A "me too" at least? Ask them if they want the same data from you. Bugs get solved faster if more people are affected.

I agree that's a good idea. I don't see any way to do it, which I guess means I need to login to leave a comment. To do that I would need an account, which I don't have. I tried to create one a while ago but it didn't work and I couldn't get a sensible answer out of Novell support. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

2008/3/11, Dave Howorth :

I wrote:

...

I just started getting this message "No space left on device" although df indicates there is about 40 GB free on the device. It's a reiser file system on a Suse 10.3 system (the file system was originally created under 9.2 if that matters). Does anybody know what this means?

I see there was a thread in October started by Ciro Iriarte with a similar question http://lists.opensuse.org/opensuse/2007-10/msg03149.html and that had Jeff Mahoney posting a patched kernel http://lists.opensuse.org/opensuse/2007-10/msg03901.html

So that leads me to some more questions :)

(1) Am I seeing the same problem? (how do I know? - reiserfstune o/p is below if it is relevant)

(2) Did Jeff's kernel fix the problem?

(3) If so, presumably the current kernel now also fixes it? (I have 2.6.22.5-31 and YaST says 2.6.22.17-0.1 is available)

(4) Am I going to regret updating to 2.6.22.17-0.1 for any other reason?

(5) I remember some recent posts about how to keep the previous kernel when updating but can't immediately find it. Can anybody point me to a how-to?

Thanks, Dave

# reiserfstune /dev/mapper/storage-data reiserfstune: Journal device has not been specified. Assuming journal is on the main device (/dev/mapper/storage-data).

Current parameters:

Filesystem state: consistent

Reiserfs super block in block 16 on 0xfd01 of format 3.6 with standard journal Count of blocks on the device: 288358400 Number of bitmaps: 8800 Blocksize: 4096 Free blocks (count of blocks - used [journal, bitmaps, data, reserved] blocks): 10415433 Root block: 85599415 Filesystem is clean Tree height: 6 Hash function used to sort names: "r5" Objectid map size 4, max 972 Journal parameters: Device [0x0] Magic [0x45a9e338] Size 8193 blocks (including 1 for journal header) (first block 18) Max transaction length 1024 blocks Max batch size 900 blocks Max commit age 30 Blocks reserved by journal: 0 Fs state field: 0x0: sb_version: 2 inode generation number: 173805158 UUID: 82287e36-4584-4b13-8a81-fef184d077df LABEL: Set flags in SB: ATTRIBUTES CLEAN

Hi.... I did a major reorganization of my data, so can't reproduce the same test case, but if i'm not wrong, the problem was mainly that I couldn't fill a FS to 100%, the problem appeared having 500MB free in some filesystems and not being able to exhaust them... Now i'm running latest kernel update and can fill completely my /home as a normal user using dd, so that means my particular issue was solved... Regards, Ciro PS: Was that 5 months ago already?! -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Howorth schreef:

Carlos E. R. wrote:

...

40GB of how much? what's the total size of the partition?

1.1 TB Probably around 500,000,000 files.

Cheers, Dave

Seems like it is fragmented... -- Enjoy your time around, Oddball (Now or never...) Besturingssysteem: Linux 2.6.24.1-6-default x86_64 Current user: oddball@AMD64x2-sfn1 System: openSUSE 11.0 (x86_64) Alpha2 KDE: 4.0.2 (KDE 4.0.2) "release 8.1" -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Tuesday 2008-03-11 at 12:11 -0700, Sloan wrote:

...

Seems like it is fragmented...

Out of inodes?

No way! Not for a reiserfs, AFAIK. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFH1uvttTMYHG2NR9URAjNaAJ9PGGi65PQP64Zu4kHiF9iIRIy+2QCfZwBo dYuatg9ntxKMkukKv1Y5Ljc= =Syma -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

tabris wrote:

Maybe ReiserFS has the same problem XFS does, wherein it can only create so many inodes if left in the default 32bit mode (in XFS, the inodenumber is related to where on disc it is physically located. This causes issues past the 1TB mark, where it starts having to use 64bit inode numbers)

reiserfs doesn't have inodes. Sloan wrote:

You're correct, of course - technically speaking, there are no inodes in reiserfs. But perhaps there is something analogous to that in this situation.

I'm not aware of anything; please let me know if you know of anything specific. The limit I am aware of is 2^32 files and I'm several orders of magnitude away from that. Don Raboud wrote:

So there is about 5% free space?

df says 3%. I make it 3.6%

It is my understanding that ext3 file systems for example reserve some space for the root user (5% by default?). Do reiser file systems do something similar? Can you use that space as root?

I don't believe reiser does that but in any case I also see the problem as root. Sloan wrote:

OK, I wasn't paying close enough attention - the OP is running the kernel from the original install, never updated, so he naturally didn't get the reiserfs fix mentioned earlier -

http://lists.opensuse.org/opensuse/2007-10/msg03901.html

That's right. I am aware of Ciro's problem that is now solved but I'm also aware of another similar problem that isn't solved and so far I don't have enough details to be able to check which problem I have or whether I have something different again. I built my problem after that fix was tested but I don't know when or if it went into the default kernel. So I don't know whether that fix is in the system I'm running. Sam Clemens wrote:

Sounds like filesystem corruption... like the free-list got lost.

I'm running reiserfsck to eliminate that possibility. So far it says: reiserfsck --check started at Tue Mar 11 11:50:44 2008 ########### Replaying journal: Done. Reiserfs journal '/dev/mapper/storage-data' in blocks [18..8211]: 0 transactions replayed Checking internal tree.. finished Comparing bitmaps..finished Checking Semantic tree: /trembl/xml/releases/2006-05-16/P97842.xml Two years to go :) No errors in the log so far. Sam Clemens wrote:

What is the exact diagnostic message?

I posted that earlier, and it's the subject of the thread :) "No space left on device".

And what is the output of df -H?

The filesystem is offline because of the reiserfsck so I can't run df now. It previously said something like: /dev/mapper/storage-data 1.1T 1.1T 40G 97% /data Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Sloan wrote:

Carlos E. R. wrote:

...

The Tuesday 2008-03-11 at 12:11 -0700, Sloan wrote:

...

...

Seems like it is fragmented...

...

Out of inodes?

No way! Not for a reiserfs, AFAIK.

You're correct, of course - technically speaking, there are no inodes in reiserfs. But perhaps there is something analogous to that in this situation.

OK, I wasn't paying close enough attention - the OP is running the kernel from the original install, never updated, so he naturally didn't get the reiserfs fix mentioned earlier - http://lists.opensuse.org/opensuse/2007-10/msg03901.html Joe -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Oddball wrote:

Dave Howorth schreef:

...

Carlos E. R. wrote:

...

40GB of how much? what's the total size of the partition?

1.1 TB Probably around 500,000,000 files.

Cheers, Dave

Seems like it is fragmented...

But fragmentation wouldn't cause available blocks to be unusable...it would (at worst) cause them to be used in a way which is very inefficient (time wise) to read and write. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Ken Schneider wrote:

Carlos E. R. pecked at the keyboard and wrote:

...

The Tuesday 2008-03-11 at 13:48 -0000, Dave Howorth wrote:

...

I just started getting this message "No space left on device" although df indicates there is about 40 GB free on the device. It's a reiser file system on a Suse 10.3 system (the file system was originally created under 9.2 if that matters). Does anybody know what this means? 40GB of how much? what's the total size of the partition?

What is the total size of the files you were trying to gzip? It has to be less then the amount of free space.

I don't see how that can be correct. I was running a gzip -r and it had been happily running for some hours and had compressed a lot of files. Now if I just try to compress the next file by itself, which is only a few kB, it objects. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Carlos E. R. wrote:

Did you update the kernel and reboot, yet?

No, I'd still like to find a list of changes that suggests there was a known reiser problem that has been fixed in the current kernel. Ciro's problem sounds very much like Maciej Pilichowski's problem to my untrained brain, but Ciro says his problem has gone away whilst Maciej's is still open. And I still don't know why YaST updater is broken. I don't like changing things - especially something as fundamental as the kernel - when there are unexplained faults. Also ...

What was the result of the reiserfsck after the reboot?

The original reiserfsck is still running! It hasn't found any errors yet. But it seems a good idea to let it finish to eliminate filesystem corruption as the problem. I still have no specific evidence that changing the kernel will actually fix my problem. There's one piece of circumstantial evidence that suggests it will (Ciro) and one that suggests it won't (Maciej) but no documentary evidence or feedback from Suse. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Ciro Iriarte wrote:

This are the reiserfs references I found on the changelog: <snip>

Thanks Ciro :) I'll try updating the kernel after the fsck finishes. It's been running 26 hours now and it's reached October 2006. My best guess is it will take another 8 hours to complete. (it's a good job it doesn't do an fsck every time it boots!) Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Wednesday 2008-03-12 at 13:43 -0000, Dave Howorth wrote:

take another 8 hours to complete. (it's a good job it doesn't do an fsck every time it boots!)

It does. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFH1+gRtTMYHG2NR9URAt9xAKCBrAakQy2m+SxUnjqpH6HP2GRvugCfQQBp fORZX9Ro3c0qpI27Alb3ym8= =D4SF -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Sam Clemens wrote:

...

I'll try updating the kernel after the fsck finishes. It's been running 26 hours now and it's reached October 2006. My best guess is it will take another 8 hours to complete. (it's a good job it doesn't do an fsck every time it boots!)

Ooooh, that's VERY Slow. Failing disk drive?

I don't believe so. The drives are part of a RAID with a 3ware controller that sends me mail if anything hiccups. And its browser interface says everything is good. The reiserfsck is still running - 47 hours now - so I think I'll abort it and try the new kernel. Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Howorth wrote:

Dave Howorth wrote:

...

The reiserfsck is still running - 47 hours now - so I think I'll abort it and try the new kernel.

Just a status update.

I aborted the reiserfsck - it hadn't found any errors to that point.

As an experiment, I then remounted the filesystem. I could then execute commands that were previously failing. This doesn't surprise me because it's consistent with the open bug report (AFAIK it's a stronger statement).

I then installed the new kernel and rebooted. Everything is working well so far. I'm thrashing the filesystem. Fingers crossed :)

Use tar to make a (logical) image of the filesystem onto another disk, then use mkfs.reiserfs to re-format the partition, and then restore your tar image back to the original disk.

Cheers, Dave

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Carlos E. R. schrieb:

The Thursday 2008-03-13 at 13:45 -0400, Sam Clemens wrote:

...

Use tar to make a (logical) image of the filesystem onto another disk, then use mkfs.reiserfs to re-format the partition, and then restore your tar image back to the original disk.

It a terabyte filesystem.

What about 'split' then? There must be a floppy in the system?! Ok but to be honest, my hardware dealer charges 164€ for a 1000GiB Re2 Caviar. So I don't see much problems, beside the cash of course. - -- All the best, Peter J. P-N. aedon DESIGNS http://www.hochzeitsbuch.info/ http://www.aedon.eu/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFH2kbJh8q3OtgoGAwRAtjkAJ4nW6eT3O6QX8FGzaTvvLWg9cNAJACggOL3 m4C9u7SmOhqIrpfjXZohycs= =oSLl -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

...

Ok but to be honest, my hardware dealer charges 164€ for a 1000GiB Re2 Caviar. So I don't see much problems, beside the cash of course.

I suppose an lvm or raid 0 with smaller drives is cheaper (the OP has lvm).

I doubt it. 1TB drives have plunged in price in the last 6 months. $/GB on sale can be as low as $/GB for a 500GB on sale. Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Content-ID: The Friday 2008-03-14 at 08:28 -0400, Greg Freemyer wrote: [ resend, previous send failed ]

...

I suppose an lvm or raid 0 with smaller drives is cheaper (the OP has lvm).

I doubt it. 1TB drives have plunged in price in the last 6 months. $/GB on sale can be as low as $/GB for a 500GB on sale.

SATA/300, Seagate ST3500630AS Barracuda 7200.10 500,0 GB 11/16/7200 102€ 0.20 €/GB ST3500320AS Barracuda 7200.11 500,0 GB 8,5/32/7200 104€ 0.21 €/GB ST3750640AS Barracuda 7200.10 750,0 GB 11/16/7200 149€ 0.20 €/GB ST3750330NS 750,0 GB 8,5/32/7200 217€ 0.29 €/GB ST31000340AS Barracuda 7200.11 1.000,0 GB 8,5/32/72000 229€ 0.23 €/GB http://www.alternate.es/html/shop/productListing4C.html?cat1=7&cat2=54&cat3=0&cat1=7&cat2=304&cat3=0&tgid=68&treeName=HARDWARE&Level1=Discos+duros&Level2=SATA&Level3=3%2C5+pulgadas& Two 500 GB disks (or 750 at 11ms) is the cheapest option still. By a small margin, true. Now, I wonder if two disks is faster than a single disk :-? - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFH2v0vtTMYHG2NR9URAhaMAKCCuYEAxc0yRa/IoaW/mFVYCKnKfwCfYSST BIsZyS1VhEqKGKMvGavFIfs= =EU/t -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Carlos E. R. schrieb:

...

...

It a terabyte filesystem. What about 'split' then? There must be a floppy in the system?!

you gotta be kidding :-p

You caught me...again! :-p - -- All the best, Peter J. P-N. aedon DESIGNS http://www.hochzeitsbuch.info/ http://www.aedon.eu/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFH2umah8q3OtgoGAwRAoe9AJ9uefboD4I0iBfDrtoPeCDWDmal8QCfbTjK m9e7Fo8LpeFm8P70WekDvVo= =JO7i -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Howorth wrote:

Dave Howorth wrote:

...

The reiserfsck is still running - 47 hours now - so I think I'll abort it and try the new kernel.

Just a status update.

I aborted the reiserfsck - it hadn't found any errors to that point.

As an experiment, I then remounted the filesystem. I could then execute commands that were previously failing. This doesn't surprise me because it's consistent with the open bug report (AFAIK it's a stronger statement).

I then installed the new kernel and rebooted. Everything is working well so far. I'm thrashing the filesystem. Fingers crossed :)

Use tar to make a (logical) image of the filesystem onto another disk, then use mkfs.reiserfs to re-format the partition, and then restore your tar image back to the original disk.

Cheers, Dave

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Friday 2008-03-14 at 10:58 -0000, Dave Howorth wrote:

Sam Clemens wrote:

...

Use tar to make a (logical) image of the filesystem onto another disk, then use mkfs.reiserfs to re-format the partition, and then restore your tar image back to the original disk.

Errm, why? As I said "Everything is working well so far".

I guess that if it fails again, a possible solution, rather a hack, is to recreate the filesystem. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFH2mRhtTMYHG2NR9URAhj/AJ96UzzP6qqSjNo16zLvC5wHszUAKwCdFI4D TyOeffuXoNhgLuumkV6piKk= =PWQZ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Monday 2008-03-17 at 17:23 -0400, Sam Clemens wrote:

So you want to tempt fate and be back in this same position next week?

Get the data off the drive, put it on new hardware, and get rid of the failing drive.

Unless you really have a secret desire to lose all your data.

There is no proof yet that the hardware is broken, but it wouldn't be a bad idea to make sure. I would try smartctl. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) iD8DBQFH3xvitTMYHG2NR9URAvOVAJ42nFDXxjzBqyR5MYh55fev2r12qgCfYpYH Fumaz7QgybGIkKuJZIdsXDo= =YWoK -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Sam Clemens wrote:

...

...

Get the data off the drive, put it on new hardware, and get rid of the failing drive.

Unless you really have a secret desire to lose all your data.

There is no proof yet that the hardware is broken, but it wouldn't be a bad idea to make sure. I would try smartctl.

If he's been having trouble reading from it, and it can't be attributed to the controller or the cable, then the disk is, by definition, unreliable.

Last Thursday, I started having problems indicating a failing drive -- a 500 GB Western Digital Caviar that I'd just installed in my #2 workstation at the office. Panicking, because of the importance (and quantity) of data on that drive, I bought the copy of SpinRite that I'd always intended to get. (It seemed like a good excuse, and I figured that *sometime* I'm going to need it.) After a quick 2-hour level 2 SpinRite run, the disk seemed to be rehabilitated, but the next morning when I came to work, there were problems again. Realizing that I'd just maxed out the memory in that machine (from 1.5 GB to 4 GB) plus adding a bigger disk drive, I figured that maybe I'd exceeded the capacity of the OEM case fan to keep things cool inside, so I took off the side panel and ran SpinRite again -- this time, the 24-hour level 4 session. And I ordered a 70+ CFM fan from NewEgg.com. Everything seemed fine today, so I thought I'd try putting the side panel back on, and instantly things went screwy again. Drive D: disappeared. Turns out, with the new SATA drive installed in the normal way, and with an ordinary SATA cable (straight), the side panel on the case bends the end of the cable over sideways where it enters the back of the drive. Removing the case panel again, and re-inserting the cable after straightening out the end of it, solved the problem. I now have drive D: back again. And I've ordered a 90-degree SATA cable from NewEgg. Maybe something like that could be going on here? Jerry in Bothell, WA -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Sam Clemens wrote:

And as for lack of fan problems...doesn't overheating damage those very same disk drives?

Of course it can. That's why I suspected a possible heat problem in the first place. Instead, it turned out just to be a smashed cable connection, and I've ended up with a copy of SpinRite that I don't need ($90) and a fan I don't need ($15) on the way. It would have been a lot cheaper if I'd carefully examined the cable in the first place -- that's why I mentioned it here. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Tue, Mar 18, 2008 at 5:51 PM, Anders Johansson wrote:

Jerry Houston wrote:

...

Last Thursday, I started having problems indicating a failing drive -- a 500 GB Western Digital Caviar that I'd just installed in my #2 workstation at the office. Panicking, because of the importance (and quantity) of data on that drive, I bought the copy of SpinRite that I'd always intended to get. (It seemed like a good excuse, and I figured that *sometime* I'm going to need it.)

SpinRite is completely bogus, your money was wasted. The advertising for it is complete and utter technobabble.

There is no software product that can save failing hardware, and you would do better not to give your money to people who claim there are. It is the technological equivalent of healing crystals

Anders

Anders, it has been years since I've looked into SpinRite, but it used to be a good product 20+ years ago. Your comment got me curious. So I went and read its data recovery section on the web: http://www.grc.com/srrecovery.htm That actually makes a lot more sense than you are giving it credit for. It is all about making a really hard effort to recover data from a damaged sector and then using reallocation to replace it. If you don't know, a typical sector read will fail if the internal checksum does not match what is expected. There is a low-level read the asks for the entire low-level sector (572 bytes?) without concern about the checksum. By calling that repeatedly, SpinRite conceivably could figure out what the correct data for the sector is. And then mark the sector for reallocation to one of the reserved sectors, and then populate the sector with the recovered data. FYI: As of very recently (Feb08) hdparm may have some of they core features needed to write something like this for Linux. But per the latest hdparm man page the low level read/write commands are typically not available for sectors beyond 128GiB. Apparently some very recent new drives do support a new command that allows low-level sector addressing beyond 128GiB. So SpinRite is likely limited to doing the extensive sector recovery to the first 128 GiB of sectors. Greg -- Greg Freemyer Litigation Triage Solutions Specialist http://www.linkedin.com/in/gregfreemyer First 99 Days Litigation White Paper - http://www.norcrossgroup.com/forms/whitepapers/99%20Days%20whitepaper.pdf The Norcross Group The Intersection of Evidence & Technology http://www.norcrossgroup.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Tuesday 18 March 2008 16:01, Anders Johansson wrote:

...

http://www.grc.com/files/sr5_lit.pdf

"SpinRite scrubs surfaces, fixing and finding any problems it encounters"

I haven't seen a disk yet, no matter how new and pristine, that couldn't be improved with a bit of surface scrubbing!

...

Sure, I use dd_rescue. It's all the other claptrap that convinces me that Gibson doesn't know what he's talking about

Maybe we should introduce him to Sam Clemens (the latter-day one, not the original...)

Anders

RRS -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Tuesday 18 March 2008 16:53, James Knott wrote:

Randall R Schulz wrote:

...

Maybe we should introduce him to Sam Clemens (the latter-day one, not the original...)

How do you know he's not the original? ;-)

Oh, yeah! Rumors of his death were _extremely_ exaggerated! RRS -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Sam Clemens wrote:

...

...

Use tar to make a (logical) image of the filesystem onto another disk, then use mkfs.reiserfs to re-format the partition, and then restore your tar image back to the original disk.

Sam Clemens wrote:

So you want to tempt fate and be back in this same position next week?

Get the data off the drive, put it on new hardware, and get rid of the failing drive.

Unless you really have a secret desire to lose all your data.

Sam Clemens wrote:

If he's been having trouble reading from it, and it can't be attributed to the controller or the cable, then the disk is, by definition, unreliable.

Sam, I know you're trying to be helpful but I believe my problem is fixed and I think you misunderstand the problem that I had. I agree with the logic in your last comment but the fact is that I did *not* have any trouble reading data. There's been no indication of any hardware problems in drives, cables, controller, motherboard, processor or anything else. The trouble I had was an inability to *write* data and that has been identified as a kernel bug and fixed by updating the kernel. My problem is solved, so let's bring the thread to a close. Please reread my postings if you still have questions about the fault. FWIW, I have previously had two drives fail in this array. The 3ware hardware has coped as it is designed to. It notifies me of a failing drive; I pull the drive and replace it with a new one (which I can get from the stock we keep) and tell the controller to rebuild the array. In case that doesn't work, there's a nightly backup :) And yes, Carlos, smart is running on the drives, monitored by the controller :) Cheers, Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Howorth wrote:

Ken Schneider wrote:

...

Carlos E. R. pecked at the keyboard and wrote:

...

The Tuesday 2008-03-11 at 13:48 -0000, Dave Howorth wrote:

...

I just started getting this message "No space left on device" although df indicates there is about 40 GB free on the device. It's a reiser file system on a Suse 10.3 system (the file system was originally created under 9.2 if that matters). Does anybody know what this means? 40GB of how much? what's the total size of the partition? What is the total size of the files you were trying to gzip? It has to be less then the amount of free space.

I don't see how that can be correct. I was running a gzip -r and it had been happily running for some hours and had compressed a lot of files. Now if I just try to compress the next file by itself, which is only a few kB, it objects.

Sounds like filesystem corruption... like the free-list got lost. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org