Hi guys,The raid mirror set of our (co-located) production server is now badly broken. Luckily, I can fix it probably remotely using the IPMI module and a backup hard disk with a full system on it. But this should not have happened!
I'm still investigating what went wrong. But I would like to announce it right now to avoid other people getting into the same or worse trouble. The package dmraid-1.0.0.rc13-14.el5_2.1.x86_64.rpm caused my raid set to be no longer activated during boot.
Here is a log from a number of versions of dmraid in RHEL5 and Heinz' directory at redhat.com. Commands are taken from mkinitrd script.
### Prepare [root@beast dmraid]# ls /dev/mapper/ control[root@beast dmraid]# /sbin/dmraid.static -ay -i -p -t 2>/dev/null | \ egrep -iv "^no " | awk -F ':' '{ print $1 }'
ddf1_EMCODEV_BEAST ### dmraid-1.0.0.rc13-4.el5.x86_64.rpm[root@beast dmraid]# ./dmraid.static_rc13-4.el5.x86_64 -ay -i -p \ "ddf1_EMCODEV_BEAST"
Segmentation fault [root@beast dmraid]# ./dmraid.static_rc13-4.el5.x86_64 -ay -i -p [root@beast dmraid]# ls /dev/mapper/ control ddf1_EMCODEV_BEAST [root@beast dmraid]# ./dmraid.static_rc13-4.el5.x86_64 -an -i -p RAID set "ddf1_EMCODEV_BEAST(null)1" is not active RAID set "ddf1_EMCODEV_BEAST(null)2" is not active RAID set "ddf1_EMCODEV_BEAST(null)3" is not active RAID set "ddf1_EMCODEV_BEAST(null)4" is not active ### dmraid-1.0.0.rc13-9.el5.src.rpm[root@beast dmraid]# ./dmraid.static_rc13-9.el5.x86_64 -ay -i -p \ "ddf1_EMCODEV_BEAST"
No RAID sets and with names: "ddf1_EMCODEV_BEAST" [root@beast dmraid]# ./dmraid.static_rc13-9.el5.x86_64 -ay -i -p [root@beast dmraid]# ls /dev/mapper control ddf1_EMCODEV_BEAST [root@beast dmraid]# ./dmraid.static_rc13-9.el5.x86_64 -an -i -p RAID set "ddf1_EMCODEV_BEAST(null)1" is not active RAID set "ddf1_EMCODEV_BEAST(null)2" is not active RAID set "ddf1_EMCODEV_BEAST(null)3" is not active RAID set "ddf1_EMCODEV_BEAST(null)4" is not active [root@beast dmraid]# ls /dev/mapper/ control ddf1_EMCODEV_BEAST [root@beast dmraid]# ./dmraid.static_rc13-9.el5.x86_64 -an -i RAID set "ddf1_EMCODEV_BEAST(null)1" is not active RAID set "ddf1_EMCODEV_BEAST(null)2" is not active RAID set "ddf1_EMCODEV_BEAST(null)3" is not active RAID set "ddf1_EMCODEV_BEAST(null)4" is not active [root@beast dmraid]# ls /dev/mapper/ control ddf1_EMCODEV_BEAST (deactivated raid set with dmraid.static_rc13-4.el5.x86_64) ### dmraid-1.0.0.rc13-14.el5_2.1.x86_64.rpm[root@beast dmraid]# ./dmraid.static_rc13-14.el5_2.1 -ay -i -p \ "ddf1_EMCODEV_BEAST"
RAID set "ddf1_EMCODEV_BEAST" was not activated [root@beast dmraid]# ls /dev/mapper control [root@beast dmraid]# ./dmraid.static_rc13-14.el5_2.1 -ay -i -p RAID set "ddf1_EMCODEV_BEAST" was not activated [root@beast dmraid]# ls /dev/mapper control ### dmraid-1.0.0.rc15-1.fc9.x86_64.rpm[root@beast dmraid]# ./dmraid.static_rc15-1.fc9.x86_64 -ay -i -p "ddf1_EMCODEV_BEAST"
RAID set "ddf1_EMCODEV_BEAST" was not activated [root@beast dmraid]# ./dmraid.static_rc15-1.fc9.x86_64 -ay -i -p RAID set "ddf1_EMCODEV_BEAST" was not activated [root@beast dmraid]# ls /dev/mapper/ controlConclusion: none of these 4 versions work like they should. The dmraid-1.0.0.rc13-9.el5.x86_64.rpm was workable with a patch to mkinitrd to call dmraid without raid set name.
Question: when calling "dmraid -an", shouldn't the -p command cause that it does not attempt to deactivate partitions?
Attached is tar of dmraid -rD output. Please take a look at it. Regards, Bas.
Attachment:
ddf1-EMCODEV_BEAST.tar.bz2
Description: application/bzip
_______________________________________________ Ataraid-list mailing list Ataraid-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/ataraid-list
