Bugzilla – Bug 548132
YaST2 automatic repair from DVD still fails to recognize RAID1 array; 11.2 unable to recognize 11.1 raid filesystem
Last modified: 2009-12-30 13:03:49 UTC
User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_8; en-us) AppleWebKit/531.9 (KHTML, like Gecko) Version/4.0.3 Safari/531.9 I thought this problem in 11.1 may have related to 11.1 RAID/Intel bugs which have been fixed. Repair fails to recognize that drives are part of RAID and wants to repair filesystem on individual drives. When I let 11.1 repair do this in the past, it caused endless booting problems requiring reinstall. Problem appears persistent. Reproducible: Always Steps to Reproduce: 1. Run automatic repair fron DVD with RAID 1 array present. 2. Follow automatic repair suggestions. 3. Actual Results: on 11.1, unbootable system. Not game to let YaST2 repair ruin my 11.2-RC2 system. Expected Results: Fixed any drive errors
YaST2 repair fails to repair brand new ext4 partition after crash.
(cont) System crashed (running Kaffeine). Running automatic repair from DVD boot after reboot. Reports damaged (brand new) ext4 partition and wants to repair. Repair fails in endless loop via repair button back to dialog box "Error detected...partition corrupted".
Subsequently ext4 partitions are OK. Due to automatic repair during hard disk boot?? RAID array still not recognised. YAST partitioner lists unmounted RAID. Individual devices and array have no filesystem. Unable to mount array.
Repair is not adapted to changes done in 11.2 and is possible that this problem is not fixed to release of 11.2.
Created attachment 324010 [details] 11.1 raid information raid1 fully functional
Created attachment 324011 [details] raid1 11.2rc1 raid1 filesystem not recognizable to 11.2rc1
Possibly this is similar to this bug: https://bugzilla.novell.com/show_bug.cgi?id=525282
It appers that the underlying problem could be a strange partition table created by YaST partitioner for RAID1. mount on 11.1 is unable to recognise filesystem type but manages to mount ext3 which functions normally. (YaST partitioner also correctly mounts.) mount: you didn't specify a filesystem type for /dev/md0 I will try type ext3 /dev/md0 on /srv type ext3 (rw) parted /dev/md0 on 11.1: (parted) check 1 Error: File system has an incompatible feature enabled. (parted) mount on 11.2 is unable mount unable to detect filesystem. parted on 11.2: # parted /dev/md0 GNU Parted 1.9.0 Using /dev/md0 Welcome to GNU Parted! Type 'help' to view a list of commands. (parted) print all Warning: /dev/md0 contains GPT signatures, indicating that it has a GPT table. However, it does not have a valid fake msdos partition table, as it should. Perhaps it was corrupted -- possibly by a program that doesn't understand GPT partition tables. Or perhaps you deleted the GPT table, and are now using an msdos partition table. Is this a GPT partition table? parted: invalid token: all Yes/No? Y (parted)
I check code and find, that repair should not try to repair md RAID. It doesn't know it. So it should not touch your system if there is not problem in raid detection. But I find that this bug report doesn't contain logs from repair and without it I cannot find what is going wrong
So could you provide y2logs? http://en.opensuse.org/Bugs/YaST#I_reported_a_YaST2_bug.2C_and_now_I_am_asked_to_.22attach_y2logs.22._What_does_that_mean.2C_and_how_do_I_do_that.3F thanks
Created attachment 325613 [details] y2logs (Have since given up and rebuilt raid array in 11.2.) Note the problem was easy to replicate - build a new raid1 array from new drives using YsT partitioner 11.1 assigning whole drive. Try to mount using YaST pertitioner in 11.2 system.
Arvin - it looks like related to some chagnes in yast2-starage. Can you provide any information what is changed in raids between 11.1 and 11.2?
The software RAID md0 contains a (GPT) partition table. Likely simply a leftover from one of the disks in the RAID. The kernel in 11.1 did not have support for that. YaST in 11.2 does not support this. As a workaround delete the partition table on the RAID with dd.
I believe the erroneous partition table was built by YaST partitioner 11.1 from clean drives after a BIOS format, on more than one occasion. mdadm happily administers a healthy RAID1, which YaST in 11.1 and 11.2RC1 builds a new GPT partition table for. Is that what is supposed to happen? If one is not meant to use a GPT partition table on a RAID1, then that option should be unavailable.
I do not think that YaST created the partition table on md with 11.1. Please provide YaST logs of such an event.
Created attachment 325943 [details] y2logs11.1 Being a linux/suse newb, I don't know enough to create a partition table from scratch. So I used the automated yast tools. Initially, the RAID1 was build using BIOS + dmraid via Yast. After problems, the RAID drives were BIOS formatted and software RAID1 built using mdraid via Yast. Have certainly done a lot of poking round trying to clarify the exact problem trying to preserve what (negligible) data I had. But when nothing could be made to work manually, have resorted to BIOS format and clean partition/rebuild. RAID1 has now been rebuilt using 11.2RC1, so I can only recover what has been logged in 11.1 prior to the rebuild.
YaST does not create a partition table in the logs from comment #17. As I tried to explain in comment #13 the partition table was likely already present on one of the disks. Also the logs seem to be incomplete. Once the RAID is made out of sdb1 and sdc1 (partitions on disks) and once out of sdb and sdc (the whole disks). But there is also one log where YaST creates a software RAID (on sdb1 and sdc1). But I'm not going to analyse the problem further since as stated already before partitioned software RAIDs are not supported by YaST on openSUSE 11.2. Feel free to make a feature request at https://features.opensuse.org/.
Resolved? That's nice. Upgrading 11.1 to 11.2 with raid setup leaves a system in an unbootable state. I could manually mount my md devices in rescue, and see the devices (raid1 and 5) How is that a "feature" request for those SW-raid users which will end up nuking their system? What can we do to avoid a full reinstall? Will reinstalling a fresh 11.2 with raid devices cause problems?
Manually fixed the problem somehow: - boot into rescue - load raid1 and raid5 modules - load dm-mod module (LVM) - mount partitions - mount proc, sysfs, and bind /dev mount - chroot into root mount point - assure that your /proc/mdstat shows all well - regenerated initrd with raid1 (and raid5 just to be sure) - sync to disk - rebooted It seems that the raid modules were not loaded on boot. Now they are. Weird.
Comment 18 is erroneous. This has been a working FEATURE. I say that because I was using it at (64bit) 10.3, 11.0, 11.1 and now 11.2. It is as was stated in Comment 20, that when you run the rescue system, THEN and ONLY THEN does the system apparently include the RAID support. In my specific case, after getting 11.2 RC2 work work correctly (with RAID1), I was able to upgrade to 11.2 "GA" without a major glitch. Accordingly, I am marking this as reopened.
Apparently we disagree. But it's open source, we accept patches.
Support for partitioned software RAIDs has been added for openSUSE 11.3 in YaST. Should fix the problem.