[opensuse] Re: Electricity Cutoffs, EXT3 and Filesystems
On Sun, Nov 2, 2008 at 10:39 AM, Volkan YAZICI
Hi,
I had used to be a ReiserFS user ~4 years and never had even a single problem with it. It was blazingly fast and there isn't a second filesystem I know of that could challange with it in the speed of recovering at the boot after a system crash or electricity cutoff.
This year I'm obligated to administrate extra ~5 production servers and as a result of major GNU/Linux headquarters moving from ReiserFS to EXT3, I started to use EXT3 in those new servers. But unfortunately, after every electricity cutoff[1], EXT3 just crashes and waits prompt from me standing at boot. I start the servers with Knoppix (Gee!) and run e2fsck on every single partition. (Keep on imagining this PITA!) No, pressing `Y' to run a fsck on the partitions doesn't work. I tried my luck with XFS, but it resulted same as EXT3.
Please, I don't want to start a flamewar between filesystems. But could anybody give any recommendations to me? Should I switch back to ReiserFS for my own mental sake? Is it possible to configure EXT3 to behave in a more automatized manner and recover from crashes with minimum human interruption? Do others also experience similar problems?
Regards.
[1] Yes, we have couples of UPS boxes around, but they are not capable of standing the load for many hours. And yes, this is "Banana Republic" and companies cut your electricity off without a clue.
-- To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Hi Can't the systems shut down when the UPS is doing it's job? They should have enogh juice to run for a minute, or the UPS set is far below spec. Neil -- There are three kinds of people: Those who can count, and those who cannot count ----------------------------------------------------------------------- ** Hi! I'm a signature virus! Copy me into your signature, please! ** ----------------------------------------------------------------------- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Volkan YAZICI
Please, I don't want to start a flamewar between filesystems. But could anybody give any recommendations to me?
I used to operate a whole bunch of Solaris systems on survey ships, and power-loss situations were always fraught. My thoughts: ReiserFS is still available and stable, why not use it for a year and see what happens? My other thought: Hardly a recommendation, because I have not tried it, but have you looked at http://jfs.sourceforge.net/ ? You could put it on an old box and pull the power cord half a dozen times to see what happens Passing thought: I've worked with instruments that use real-time linux, and most of those had a crash-proof file system of one sort or another. A bit of research into what they use might be good. I do know that Arcom use what they call "Compressed Journalling Flash File System (JFFS2)", but there are larger real-time systems with hard disks. One data acquisition system I used just used ext3 but mounted nosync. It didn't cause much trouble. Neil wrote:
Can't the systems shut down when the UPS is doing it's job? They should have enogh juice to run for a minute, or the UPS set is far below spec.
It is the obvious solution: rig the serial output from the UPS to trigger a shutdown immediately. It is possible hough that the UPS may not be large enough to provide the amps required for all those servers during normal operation. I too have worked in Banana republics and on ships where the UPS is more trouble than it is worth. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Robert E A Harvey wrote:
My other thought: Hardly a recommendation, because I have not tried it, but have you looked at http://jfs.sourceforge.net/ ? You could put it on an old box and pull the power cord half a dozen times to see what happens
There is, to my knowledge, no filesystem in existence with 100% immunity to power outages. JFS is as good or as bad as any other in that respect. The OPs problems seems to be that his systems are not setup for running an automatic fsck at startup. Only if the initial fsck fails will the system refuse to continue - it should not just always open a prompt and expect a manual fsck. /Per -- /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday, 2008-11-03 at 10:15 +0100, Per Jessen wrote:
My other thought: Hardly a recommendation, because I have not tried it, but have you looked at http://jfs.sourceforge.net/ ? You could put it on an old box and pull the power cord half a dozen times to see what happens
There is, to my knowledge, no filesystem in existence with 100% immunity to power outages. JFS is as good or as bad as any other in that respect.
I think reiserfs is more resilient in this situation. At lest, it checks faster.
The OPs problems seems to be that his systems are not setup for running an automatic fsck at startup. Only if the initial fsck fails will the system refuse to continue - it should not just always open a prompt and expect a manual fsck.
Maybe the script trigering the automatic fsck fails for some other reason, like a faulty line in fstab or a not present partition. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkkO5TEACgkQtTMYHG2NR9XQ7wCdG0slw1mb7pcWh435e4dvfsLN x7EAoJlQ/ysTZL8wR+FuFOo954BI/rGM =9kVm -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Carlos E. R. wrote:
On Monday, 2008-11-03 at 10:15 +0100, Per Jessen wrote:
My other thought: Hardly a recommendation, because I have not tried it, but have you looked at http://jfs.sourceforge.net/ ? You could put it on an old box and pull the power cord half a dozen times to see what happens
There is, to my knowledge, no filesystem in existence with 100% immunity to power outages. JFS is as good or as bad as any other in that respect.
I think reiserfs is more resilient in this situation. At lest, it checks faster.
I can't compare them, but it seems to me that no filesystem can guarantee clearing the appropriate "dirty" bit when power is lost at precisely the wrong nanosecond. I'm no expert though.
The OPs problems seems to be that his systems are not setup for running an automatic fsck at startup. Only if the initial fsck fails will the system refuse to continue - it should not just always open a prompt and expect a manual fsck.
Maybe the script trigering the automatic fsck fails for some other reason, like a faulty line in fstab or a not present partition.
Yep, my thoughts exactly. -- /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday, 2008-11-03 at 13:24 +0100, Per Jessen wrote:
I think reiserfs is more resilient in this situation. At lest, it checks faster.
I can't compare them, but it seems to me that no filesystem can guarantee clearing the appropriate "dirty" bit when power is lost at precisely the wrong nanosecond. I'm no expert though.
Absolutely, there is no guarantee. And when reiserfs really fails, it fails royally. However, I understand it was a design goal, to survive power-cuts, and seems to work quite well in this respect, with some quirks. Sometimes the self check thinks everything is ok and it is not: then you need an fsck made from the rescue dvd. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkkPJFMACgkQtTMYHG2NR9WIqACffrC0/WYq1w7ZIsORgTgmfmdn fAsAn1+c4i9Mesr3/ELLzfmEnUwdqiq3 =9fh9 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Carlos E. R. wrote:
On Monday, 2008-11-03 at 13:24 +0100, Per Jessen wrote:
I think reiserfs is more resilient in this situation. At lest, it checks faster.
I can't compare them, but it seems to me that no filesystem can guarantee clearing the appropriate "dirty" bit when power is lost at precisely the wrong nanosecond. I'm no expert though.
Absolutely, there is no guarantee. And when reiserfs really fails, it fails royally.
That is why, several years ago, I bought a duplicate of my primary drive and every major change gets rsync'ed to my backup drive Oddly enough, I have had great luck with reiserfs and just fair luck with ext3. -- Duaine Hechler Piano, Player Piano, Pump Organ Tuning, Servicing & Rebuilding Reed Organ Society Member Florissant, MO 63034 (314) 838-5587 dahechler@charter.net www.hechlerpianoandorgan.com -- Home & Business user of Linux - 10 years -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Mon, Nov 3, 2008 at 11:18 AM, Carlos E. R.
I think reiserfs is more resilient in this situation. At lest, it checks faster. I can't compare them, but it seems to me that no filesystem can guarantee clearing the appropriate "dirty" bit when power is lost at precisely the wrong nanosecond. I'm no expert though. Absolutely, there is no guarantee. And when reiserfs really fails, it fails royally.
The only time I ever had reiser fail was when I got a bad sector in the beginning of a partition that screwed up the directory tables. I was able to use dd_rescue and repair it tho to save the data.
However, I understand it was a design goal, to survive power-cuts, and seems to work quite well in this respect, with some quirks. Sometimes the self check thinks everything is ok and it is not: then you need an fsck made from the rescue dvd.
I switched to reiser back when SuSE did in like 7.3 or so. I got tired of ext2's inability to recover from a bad poweroff. I've used reiser exclusively since, and will continue to do so. It recovers fast and it's performance is fast. Shame that people crap on it because the designer is a murderer. It's still one of the best file systems, and it is still being maintained and extended without him. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Larry Stotler escribió: Shame that people crap on it
because the designer is a murderer.
You are not well informed then, Jeff Mahoney published the real reasons behind this change, 2 years ago http://lists.opensuse.org/opensuse-factory/2006-09/msg00542.html It has **nothing** to do with the terrible personal problems of Hans Reiser.
and extended without him.
That's not quite right. -- "Good, Fast, Cheap: Pick any two (you can't have all three)." Cristian Rodríguez R. Platform/OpenSUSE - Core Services SUSE LINUX Products GmbH Research & Development http://www.opensuse.org/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Content-ID:
Larry Stotler escribió: Shame that people crap on it
because the designer is a murderer.
You are not well informed then, Jeff Mahoney published the real reasons behind this change, 2 years ago
http://lists.opensuse.org/opensuse-factory/2006-09/msg00542.html
It has **nothing** to do with the terrible personal problems of Hans Reiser.
Nobody was talking about that change. Yet, at least. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkkQGCMACgkQtTMYHG2NR9UIbACgjtkxY557+/3Hl0pyK08bVOP5 a4AAn0gSF1p7gvMRbZrhY7qRrV1emhkW =IRUR -----END PGP SIGNATURE-----
On Mon, Nov 3, 2008 at 11:34 PM, Cristian Rodríguez
Larry Stotler escribió: Shame that people crap on it
because the designer is a murderer.
You are not well informed then, Jeff Mahoney published the real reasons behind this change, 2 years ago
http://lists.opensuse.org/opensuse-factory/2006-09/msg00542.html
It has **nothing** to do with the terrible personal problems of Hans Reiser.
Yeah, I was kinda off the cuff with that one, but unfortunately, that has become the perception. I should have mentioned that a lot of the kernel devs didn't like his attitude that much either(especially the push for reiser4). However, his personal problems and attitudes have pushed a great many people away from using it as much as the M$ deal pushed people from Novell/openSUSE. It may be a small percentage, but it did happen. However, ext3 is nowhere near as good as resier3 from my testing. It doesn't recover very well from unexpected power loss, isn't as stable, and it isn't as fast as reiser3. I haven't toyed with reiser4 much, but a friend of mine has, and he says it's really fast and he hasn't had any stability problems. He's running a software raid on 3 750GB drives with amazing speed.
and extended without him.
That's not quite right.
AFAIK, there has been some work on reiser4, but not enough. Yes, moving to resier4 requires a reformat, but so does moving to any other file system. ext4 has shown a lot of promise, but it's still not up to speed from what I have seen. If they can get ext4 up to speed, I'll switch since reiser doesn't look like it has a long term solution, but until then, I will keep using it. I recently mounted a resier3 partition from 11.0 under SuSE 7.3 with no problems. That shows a great deal of backwards compatibility to say the least. I don't reformat my data partitions, and some are several years old with no problems. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Carlos E. R. wrote:
And when reiserfs really fails, it fails royally.
If something is worth doing, it's worth doing well. ;-) -- Use OpenOffice.org http://www.openoffice.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Sun, Nov 2, 2008 at 10:39 AM, Volkan YAZICI
Hi,
I had used to be a ReiserFS user ~4 years and never had even a single problem with it. It was blazingly fast and there isn't a second filesystem I know of that could challange with it in the speed of recovering at the boot after a system crash or electricity cutoff.
This year I'm obligated to administrate extra ~5 production servers and as a result of major GNU/Linux headquarters moving from ReiserFS to EXT3, I started to use EXT3 in those new servers. But unfortunately, after every electricity cutoff[1], EXT3 just crashes and waits prompt from me standing at boot. I start the servers with Knoppix (Gee!) and run e2fsck on every single partition. (Keep on imagining this PITA!) No, pressing `Y' to run a fsck on the partitions doesn't work. I tried my luck with XFS, but it resulted same as EXT3.
Please, I don't want to start a flamewar between filesystems. But could anybody give any recommendations to me? Should I switch back to ReiserFS for my own mental sake? Is it possible to configure EXT3 to behave in a more automatized manner and recover from crashes with minimum human interruption? Do others also experience similar problems?
Regards.
[1] Yes, we have couples of UPS boxes around, but they are not capable of standing the load for many hours. And yes, this is "Banana Republic" and companies cut your electricity off without a clue.
--
I stopped using ext3 for root because of its bad crash recovery and its slowness (I don't know if this has improved) and changed to xfs which has performed well under power failure and crash conditions. Regards Dave P
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Neil wrote:
On Sun, Nov 2, 2008 at 10:39 AM, Volkan YAZICI
wrote: I had used to be a ReiserFS user ~4 years and never had even a single problem with it. ...
[1] Yes, we have couples of UPS boxes around, but they are not capable of standing the load for many hours. And yes, this is "Banana Republic" and companies cut your electricity off without a clue. ... -- To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Can't the systems shut down when the UPS is doing it's job? They should have enogh juice to run for a minute, or the UPS set is far below spec.
Errm, Neil, you do realize that the original post was on the debian list not the suse list I hope? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Monday, 2008-11-03 at 16:56 -0000, Dave Howorth wrote:
Errm, Neil, you do realize that the original post was on the debian list not the suse list I hope?
Wow. Which means that the OP will have seen nothing of our answers, and that the problem he had with getting the prompt instead of autofsck is something peculiar to Debian. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkkPq9MACgkQtTMYHG2NR9WKQwCgl5JzCfCnzzkQcu6gKPzss0xp jJsAn0UBc0v3ZNcpwhndKBpWXCyFGUo8 =UgTt -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (10)
-
Carlos E. R.
-
Cristian Rodríguez
-
Dave Howorth
-
Dave Plater
-
Duaine & Laura Hechler
-
James Knott
-
Larry Stotler
-
Neil
-
Per Jessen
-
Robert E A Harvey