Conclusion: DO NOT USE SOFTWARE RAID-1 WITH EXT3 I've started with this conclusion cause i am positive that raid-1 and ext3 are causing data corruption. I've upgraded from 7.1 to 7.2 and from the first date i run into data corruption problems @ least once per day (i was monitoring once per day only) after 7 days we bought a brand new server cause the previous was very old and i thought that the problems where caused of the hard disks (avg of 5 years old ata/33). After moving everything to the new server data corruption was something happened every 2-3 hours, rebooting waiting about 20-30 minutes and data corruption was visible on database files After a lot of search in the net i come in redhat forums and a read a post regarding similar problems. The poster switched back to ext2 to solve his problems as i did, after 24 hours the system is stable and i do not have any problems. Just wanted to post this message so any one other having the same problems on production system should now what to do and not spending 15 days trying to figure out what is causing the problems and even thinking switching to windows os :-( Regards Nick Ameladiotis Help authoring tools http://www.visagesoft.com Shareware Registration Services http://www.v-share.com
* Visage Tech Support (support@v-share.com) [20020307 19:04]:> Conclusion: > DO NOT USE SOFTWARE RAID-1 WITH EXT3> I've started with this conclusion cause i am positive that > raid-1 and ext3 are causing data corruption.Our main university mail servers are chewing mails with RAID1+ext3 configuration for quite some time now. Absolutely zero problems with locally compiled mainline 2.4 kernels. Just my 2 eurocents. (knock-knock-knock) Running Debian woody. Peter -- .+'''+. .+'''+. .+'''+. .+'''+. .+'' Kelemen Péter / \ / \ / fuji@elte.hu .+' `+...+' `+...+' `+...+' `+...+'
I've been using it for months without any such problems. I suspect many others have, too. Sound like you're jumping to conclusions, unless it is a specific version with a problem, I suppose. On Thu, Mar 07, 2002 at 07:04:46PM +0200, Visage Tech Support wrote:> Conclusion: > DO NOT USE SOFTWARE RAID-1 WITH EXT3
thanks for telling me, I will keep that in mind since i was just about to setup a system like that. Any idea what is at fault? On Thu, 7 Mar 2002, Visage Tech Support wrote:> Conclusion: > DO NOT USE SOFTWARE RAID-1 WITH EXT3 > > I've started with this conclusion cause i am positive that raid-1 and ext3 > are causing data corruption. > I've upgraded from 7.1 to 7.2 and from the first date i run into data > corruption problems @ least once per day (i was monitoring once per day > only) after 7 days we bought a brand new server cause the previous was very > old and i thought that the problems where caused of the hard disks (avg of 5 > years old ata/33). > After moving everything to the new server data corruption was something > happened every 2-3 hours, rebooting waiting about 20-30 minutes and data > corruption was visible on database files > > After a lot of search in the net i come in redhat forums and a read a post > regarding similar problems. > The poster switched back to ext2 to solve his problems as i did, after 24 > hours the system is stable and i do not have any problems. > Just wanted to post this message so any one other having the same problems > on production system should now what to do and not spending 15 days trying > to figure out what is causing the problems and even thinking switching to > windows os :-( > > Regards > Nick Ameladiotis > Help authoring tools > http://www.visagesoft.com > Shareware Registration Services > http://www.v-share.com > > > > > _______________________________________________ > Ext3-users mailing list > Ext3-users@redhat.com > https://listman.redhat.com/mailman/listinfo/ext3-users >
Right now, 2.4.19-pre1-ac2. But, as said, been using it for quite a while, mostly with -ac, and for a while with AM's patches. Don't remember the first kernel used. This system is mirroring 2 60G IDE drives, both masters on one controller. There is also a slave 120G IDE used for nightly backups, and a slave CD/R used very rarely. Two ext3 on s/w raid1: md0 is root, ~6G; md1 is ~50G. swap space accounts for the rest, not part of the raid (don't care if it crashes, just that I don't lose the data being mirrored, this one is just a workstation). All file systems right now are ext3 on all systems, except one still running 2.2.x. Most are on s/w raid1. On Thu, Mar 07, 2002 at 10:31:45PM +0200, Visage Tech Support wrote:> Could you please specify the kernel that you are using ? > And also the filesystem sizes ? > > I have tried both 2.4.7-10 and 2.4.9-31 kernels with same results corruption > every 10-20 minutes
On Thu, Mar 07, 2002 at 07:04:46PM +0200, Visage Tech Support wrote:> Conclusion: > DO NOT USE SOFTWARE RAID-1 WITH EXT3 > > I've started with this conclusion cause i am positive that raid-1 and ext3 > are causing data corruption.Running 120GB Raid 1 with ext3 on 2 IDE 60GB Maxtor drives for a year now - No problems yet ... Root-FS on SCSI though Flo -- Florian Lohoff flo@rfc822.org +49-5201-669912 Nine nineth on september the 9th Welcome to the new billenium
Alle 18:04, giovedì 7 marzo 2002, hai scritto:> Conclusion: > DO NOT USE SOFTWARE RAID-1 WITH EXT3 >RedHat 7.2 - 2x HD 18GB SCSI ( IBM-PSG Model: ST318404LW ), Controller: <Adaptec 29160B Ultra160 SCSI adapter> - raid 1 ( software ) - lvm on top raid [ghigo@dolly ghigo]$ mount | grep ext3 /dev/vg_moscow/lv_root on / type ext3 (rw,quota) /dev/vg_moscow/lv_data on /mnt/data type ext3 (rw,quota) Work fine for 2 month now, with postgresql, apache, tomcat
why would you want to use a LVM on top of RAID? On Fri, 8 Mar 2002, Goffredo Baroncelli wrote:> Alle 18:04, gioved 7 marzo 2002, hai scritto: > > Conclusion: > > DO NOT USE SOFTWARE RAID-1 WITH EXT3 > > > RedHat 7.2 > - 2x HD 18GB SCSI ( IBM-PSG Model: ST318404LW ), Controller: <Adaptec > 29160B Ultra160 SCSI adapter> > - raid 1 ( software ) > - lvm on top raid > > [ghigo@dolly ghigo]$ mount | grep ext3 > /dev/vg_moscow/lv_root on / type ext3 (rw,quota) > /dev/vg_moscow/lv_data on /mnt/data type ext3 (rw,quota) > > Work fine for 2 month now, with postgresql, apache, tomcat > > > > _______________________________________________ > Ext3-users mailing list > Ext3-users@redhat.com > https://listman.redhat.com/mailman/listinfo/ext3-users >
Just wanted to inform everyone that i was wrong, the problem was caused of faulty memory, it is true a brand new hardware with brand new faulty memory this is why my first thoughts was about ext3 and raid-1, the corrupted files where reproduced also after switching to ext2 and after the first 48 hours. I had to run memtest86 to find out that my memory was faulty, purchase new mem faulty also, replace the last memory and everything worked just fine. My apologies for driving some of you to the wrong direction, i am also going to enable ext3 if i run into problems again then.....we'll see Regards Nick Ameladiotis Help authoring tools http://www.visagesoft.com Shareware Registration Services http://www.v-share.com ----- Original Message ----- From: "Bill Rugolsky Jr." <rugolsky@ead.dsa.com> To: "Visage Tech Support" <support@v-share.com> Sent: Thursday, March 07, 2002 11:05 PM Subject: Re: DO NOT USE Software Raid1 and Ext3> On Thu, Mar 07, 2002 at 10:30:10PM +0200, Visage Tech Support wrote: > > I have tried kernel 2.4.7-10 (the one shipped with redhat cdrom) andalso> > 2.4.9-31 with the same results. > > As for every other sttings the defaults of 7.2 installation since i alsohad> > to apply a fresh install on the new server > > > > I have created one partition of 50g on both raid disks, could this huge > > partition causing the problems ? > > In my previous hardware the partition size was 10G so i assume thatpartions> > is not the problem but buggy implementation of raid and ext3 i cannotfigure> > anything else after 15 days of trying to have a stable system > > Many of us are using it without issue, so the first guess is that it is > a poor hardware interaction. If you have uncovered a serious softwarebug, we> want you to describe it in detail, so it can be found and killed. > > What hardware? > > Motherboard? > RAM? > SCSI or IDE? > Disk Controller? > Disks > Configuration? Do you have more than one IDE drive > on a channel? > Video controller > > What is the symptom of corruption? > > Messages in /var/log/syslog? > An Oops? Please post the decoded Oops. > Errors in fsck on reboot? What are the errors? > > Which applications are you running? What's the load like? > > > Where to find relevant info: > > cmds: > uname -a > dmesg > lspci -vv > top > pstree > > Files: > > /proc/cmdline > /proc/cpuinfo > /etc/modules.conf > /proc/scsi/scsi > /proc/scsi/*/[0-9]* > /proc/ide/drivers > /proc/ide/*/model > /proc/ide/*/settings > /proc/ide/*/settings > /proc/modules > /proc/interrupts > /proc/partitions > /etc/raidtab > /proc/mdstat > /etc/fstab > /proc/mounts > /var/log/syslog > > > Regards, > > Bill Rugolsky >