Hello!
On my server PC i have Centos 7 installed. CentOS Linux release 7.6.1810. There are four arrays RAID1 (software RAID) md124 - /boot/efi md125 - /boot md126 - /bd md127 - /
I have configured booting from both drives, everything works fine if both drives are connected. But if I disable any drive from which RAID1 is built the system crashes, there is a partial boot and as a result the Entering emergency mode. By all rules RAID1 in case of similar failure of one of disks the system has to recognize this moment at loading, analyze that from the array only one disk is available and forcibly assemble inactive RAID array in case it is degraded in the mode when the second disk is missing may be... In Debian system this problem does not occur, the initramfs runs the script /scripts/local-block/mdadm builds the array from one dostoinogo disc ( lack of drive is not important, it's either a drive failure or disconnection from power-random or not random). In this script You can see the following ...... if [ "$COUNT" = "$MAX" ] then # Poor man's mdadm-last-resort@.service for incremental devices mdadm -q --run /dev/md?* # And last try for all others mdadm -q --assemble --scan --run rm -f /run/count.mdadm.initrd ....
Why in server Centos 7 this problem is still not solved? Will this be fixed in upcoming releases?
Best Regards AndersonGH
On 4/3/19 5:15 AM, Konstantin Msk via CentOS wrote:
Hello!
On my server PC i have Centos 7 installed. CentOS Linux release 7.6.1810. There are four arrays RAID1 (software RAID) md124 - /boot/efi md125 - /boot md126 - /bd md127 - /
I have configured booting from both drives, everything works fine if both drives are connected. But if I disable any drive from which RAID1 is built the system crashes, there is a partial boot and as a result the Entering emergency mode. By all rules RAID1 in case of similar failure of one of disks the system has to recognize this moment at loading, analyze that from the array only one disk is available and forcibly assemble inactive RAID array in case it is degraded in the mode when the second disk is missing may be...
[please use a subject appropriate to the message.]
I believe you are hitting this bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1451660
That is, if a drive is unplugged while the system is not running, mdadm will not reassemble the array on boot.
If a drive fails while the system is running, mdadm should fail it out of the array and the system will boot properly.
Hoping for a fix in the next minor release.
On 4/3/19 2:15 AM, Konstantin Msk via CentOS wrote:
But if I disable any drive from which RAID1 is built the system crashes, there is a partial boot and as a result the Entering emergency mode.
I haven't seen that behavior in the past. Do you have any logs or console output from that event?
Hello! rdsosreport.txt file from emergency mode https://www.asuswebstorage.com/navigate/a/#/s/BBB3BEB3EFE2428BB85BF35ED38197... journalctl -xe https://www.asuswebstorage.com/navigate/a/#/s/99FBCF1467F647CEAD2087697B779A... and console https://www.asuswebstorage.com/navigate/a/#/s/E6FBFEB42CF34C31B2F533560A624A... https://www.asuswebstorage.com/navigate/a/#/s/45AE76C7B5114916BA38EF7940F2FA...
AndersonGH
Четверг, 4 апреля 2019, 2:44 +03:00 от Gordon Messmer gordon.messmer@gmail.com:
On 4/3/19 2:15 AM, Konstantin Msk via CentOS wrote:
But if I disable any drive from which RAID1 is built the system crashes, there is a partial boot and as a result the Entering emergency mode.
I haven't seen that behavior in the past. Do you have any logs or console output from that event?
CentOS mailing list CentOS@centos.org https://lists.centos.org/mailman/listinfo/centos
Right, that's my problem.
a drive is unplugged... while the system is not running mdadm will not reassemble the array on boot.
Red Hat Bugzilla – Bug 1451660 Write that Fixed In Version:dracut-033-546.el7
I have drucat version 033-554.el7 and this bag don't fixed!
I believe you are hitting this bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1451660
That is, if a drive is unplugged while the system is not running, mdadm will not reassemble the array on boot.
If a drive fails while the system is running, mdadm should fail it out of the array and the system will boot properly.
Hoping for a fix in the next minor release.
On 4/4/19 6:17 AM, Konstantin Msk via CentOS wrote:
a drive is unplugged... while the system is not running mdadm will not reassemble the array on boot.
Red Hat Bugzilla – Bug 1451660 Write that Fixed In Version:dracut-033-546.el7
That's probably a mistake. The bug isn't closed.
I have drucat version 033-554.el7 and this bag don't fixed!
One of the suggested work-arounds in that bug report is to revert to an older version of dracut and rebuild your initrd. You'd need to get:
dracut-033-535.el7.x86_64.rpm dracut-config-rescue-033-535.el7.x86_64.rpm dracut-network-033-535.el7.x86_64.rpm kexec-tools-2.0.15-13.el7.x86_64.rpm
...from the vault: http://vault.centos.org/7.5.1804/updates/x86_64/Packages/
Downgrade those packages, and then run "dracut -f" to rebuild the initrd for your kernel.
Hello everybody! I have good news, the problem is gone.
For everyone who has this problem, I did the following
I use rpm...
Get list of installed packages # rpm -qa | grep "dracut" dracut-033-554.el7.x86_64 dracut-network-033-554.el7.x86_64 dracut-config-rescue-033-554.el7.x86_64
Delete packages
# rpm -e --nodeps "dracut-config-rescue-033-554.el7.x86_64" # rpm -e --nodeps "dracut-network-033-554.el7.x86_64" # rpm -e --nodeps "dracut-033-554.el7.x86_64" I download old packages in local directory and install it # rpm -ivh dracut-033-535.el7_5.1.x86_64.rpm Preparation... ################################# [100%] Update / install... 1:dracut-033-535.el7_5.1 ################################# [100%] # rpm -ivh dracut-config-rescue-033-535.el7_5.1.x86_64.rpm Preparation... ################################# [100%] Update / install... 1:dracut-config-rescue-033-535.el7_################################# [100%] # rpm -ivh dracut-network-033-535.el7_5.1.x86_64.rpm Preparation... ################################# [100%] Update / install... 1:dracut-network-033-535.el7_5.1 ################################# [100%]
check installed... # rpm -qa | grep "dracut" dracut-033-535.el7_5.1.x86_64 dracut-network-033-535.el7_5.1.x86_64 dracut-config-rescue-033-535.el7_5.1.x86_64 After that, the system successfully booted with one disk disabled. The array was assembled at one disk . After connecting and booting with two attached disks manually add disconnected partitions array. # mdadm /dev/md124 --add /dev/sda1 mdadm: hot added /dev/sda1 # mdadm /dev/md125 --add /dev/sda2 mdadm: hot added /dev/sda2 # mdadm /dev/md126 --add /dev/sda3 mdadm: re-added /dev/sda3 # mdadm /dev/md127 --add /dev/sda4 mdadm: re-added /dev/sda4
And waiting for the end of the recovery process for all arrays # cat /proc/mdstat
Thanks for your quick help!
Best Regards AndersonGH
Четверг, 4 апреля 2019, 17:24 +03:00 от Gordon Messmer gordon.messmer@gmail.com:
On 4/4/19 6:17 AM, Konstantin Msk via CentOS wrote:
a drive is unplugged... while the system is not running mdadm will not reassemble the array on boot.
Red Hat Bugzilla – Bug 1451660 Write that Fixed In Version:dracut-033-546.el7
That's probably a mistake. The bug isn't closed.
I have drucat version 033-554.el7 and this bag don't fixed!
One of the suggested work-arounds in that bug report is to revert to an older version of dracut and rebuild your initrd. You'd need to get:
dracut-033-535.el7.x86_64.rpm dracut-config-rescue-033-535.el7.x86_64.rpm dracut-network-033-535.el7.x86_64.rpm kexec-tools-2.0.15-13.el7.x86_64.rpm
...from the vault: http://vault.centos.org/7.5.1804/updates/x86_64/Packages/
Downgrade those packages, and then run "dracut -f" to rebuild the initrd for your kernel.
CentOS mailing list CentOS@centos.org https://lists.centos.org/mailman/listinfo/centos
-
Chris Schanzle -
Gordon Messmer -
Konstantin Msk