[ovirt-users] Re: Fwd: Re: Fwd: Fwd: Issues with Gluster Domain
As far as I know, oVirt 4.4 uses gluster v7.X , so you will eventually have to upgrade the version. As I mentioned, I have created my new volume while I was running a higher version and copied the data to it, which prevented the acl bug hitting me again. I can recommend you to: 1. Mount the new gluster volume via FUSE 2. Wipe the data, as you don't need it 3. Attach the new gluster volume as a fresh storage domain in ovirt 4. Live migrate the VM disks to the new volume prior to upgrading gluster I cannot guarantee that the issue will be avoided , but it's worth trying . Best Regards, Strahil Nikolov На 23 юни 2020 г. 23:42:13 GMT+03:00, C Williams написа: >Strahil, > >Thanks for getting back with me ! > >Sounds like it is best to evacuate VM disks to another storage domain >-- >if possible from a Gluster storage domain -- prior to an upgrade . > >Per your questions ... > >1. Which version of oVirt and Gluster are you using ? >oVirt 4.3.8.2-1 >Gluster 6.5 , 6.7, 6.9 (depends on the cluster) > >2. You now have your old gluster volume attached to oVirt and the new >volume unused, right ? >Correct -- intending to dispose of data on the new volume since the old >one is now working > >3. Did you copy the contents of the old volume to the new one ? >I did prior to trying to downgrade the Gluster version on the hosts for >the old volume. I am planning to delete the data now that the old >volume is >working. > >Thanks Again For Your Help !! > >On Mon, Jun 22, 2020 at 11:34 PM Strahil Nikolov > >wrote: > >> As I told you, you could just downgrade gluster on all nodes and >later >> plan to live migrate the VM disks. >> I had to copy my data to the new volume, so I can avoid the ACL bug , >when >> I use newer versions of gluster. >> >> >> Let's clarify some details: >> 1. Which version of oVirt and Gluster are you using ? >> 2. You now have your old gluster volume attached to oVirt and the >new >> volume unused, right ? >> 3. Did you copy the contents of the old volume to the new one ? >> >> Best Regards, >> Strahil Nikolov >> >> На 23 юни 2020 г. 4:34:19 GMT+03:00, C Williams > >> написа: >> >Strahil, >> > >> >Thank You For Help ! >> > >> >Downgrading Gluster to 6.5 got the original storage domain working >> >again ! >> > >> >After, I finished my copy of the contents of the problematic volume >to >> >a >> >new volume, I did the following >> > >> >Unmounted the mount points >> >Stopped the original problematic Gluster volume >> >On each problematic peer, >> >I downgraded Gluster to 6.5 >> >(yum downgrade glusterfs-6.5-1.el7.x86_64 >> >vdsm-gluster-4.30.46-1.el7.x86_64 >> >python2-gluster-6.5-1.el7.x86_64 glusterfs-libs-6.5-1.el7.x86_64 >> >glusterfs-cli-6.5-1.el7.x86_64 glusterfs-fuse-6.5-1.el7.x86_64 >> >glusterfs-rdma-6.5-1.el7.x86_64 glusterfs-api-6.5-1.el7.x86_64 >> >glusterfs-server-6.5-1.el7.x86_64 glusterfs-events-6.5-1.el7.x86_64 >> >glusterfs-client-xlators-6.5-1.el7.x86_64 >> >glusterfs-geo-replication-6.5-1.el7.x86_64) >> >Restarted glusterd (systemctl restart glusterd) >> >Restarted the problematic Gluster volume >> >Reattached the problematic storage domain >> >Started the problematic storage domain >> > >> >Things work now. I can now run VMs and write data, copy virtual >disks, >> >move >> >virtual disks to other storage domains, etc. >> > >> >I am very thankful that the storage domain is working again ! >> > >> >How can I safely perform upgrades on Gluster ! When will it be safe >to >> >do >> >so ? >> > >> >Thank You Again For Your Help ! >> > >> >On Mon, Jun 22, 2020 at 10:58 AM C Williams > >> >wrote: >> > >> >> Strahil, >> >> >> >> I have downgraded the target. The copy from the problematic volume >to >> >the >> >> target is going on now. >> >> Once I have the data copied, I might downgrade the problematic >> >volume's >> >> Gluster to 6.5. >> >> At that point I might reattach the original ovirt domain and see >if >> >it >> >> will work again. >> >> But the copy is going on right now. >> >> >> >> Thank You For Your Help ! >> >> >> >> On Mon, Jun 22, 2020 at 10:52 AM Strahil Nikolov >> > >> >> wrote: >> >> >> >>> You should ensure that in the storage domain tab, the old >storage >> >is >> >>> not visible. >> >>> >> >>> I still wander why yoiu didn't try to downgrade first. >> >>> >> >>> Best Regards, >> >>> Strahil Nikolov >> >>> >> >>> На 22 юни 2020 г. 13:58:33 GMT+03:00, C Williams >> > >> >>> написа: >> >>> >Strahil, >> >>> > >> >>> >The GLCL3 storage domain was detached prior to attempting to add >> >the >> >>> >new >> >>> >storage domain. >> >>> > >> >>> >Should I also "Remove" it ? >> >>> > >> >>> >Thank You For Your Help ! >> >>> > >> >>> >-- Forwarded message - >> >>> >From: Strahil Nikolov >> >>> >Date: Mon, Jun 22, 2020 at 12:50 AM >> >>> >Subject: Re: [ovirt-users] Re: Fwd: Fwd: Issues with Gluster >Domain >> >>> >To: C Williams >> >>> >Cc: users >> >>> > >> >>> > >> >>> >You can't add the new volume as it contains the same d
[ovirt-users] Re: Fwd: Re: Fwd: Fwd: Issues with Gluster Domain
As I told you, you could just downgrade gluster on all nodes and later plan to live migrate the VM disks. I had to copy my data to the new volume, so I can avoid the ACL bug , when I use newer versions of gluster. Let's clarify some details: 1. Which version of oVirt and Gluster are you using ? 2. You now have your old gluster volume attached to oVirt and the new volume unused, right ? 3. Did you copy the contents of the old volume to the new one ? Best Regards, Strahil Nikolov На 23 юни 2020 г. 4:34:19 GMT+03:00, C Williams написа: >Strahil, > >Thank You For Help ! > >Downgrading Gluster to 6.5 got the original storage domain working >again ! > >After, I finished my copy of the contents of the problematic volume to >a >new volume, I did the following > >Unmounted the mount points >Stopped the original problematic Gluster volume >On each problematic peer, >I downgraded Gluster to 6.5 >(yum downgrade glusterfs-6.5-1.el7.x86_64 >vdsm-gluster-4.30.46-1.el7.x86_64 >python2-gluster-6.5-1.el7.x86_64 glusterfs-libs-6.5-1.el7.x86_64 >glusterfs-cli-6.5-1.el7.x86_64 glusterfs-fuse-6.5-1.el7.x86_64 >glusterfs-rdma-6.5-1.el7.x86_64 glusterfs-api-6.5-1.el7.x86_64 >glusterfs-server-6.5-1.el7.x86_64 glusterfs-events-6.5-1.el7.x86_64 >glusterfs-client-xlators-6.5-1.el7.x86_64 >glusterfs-geo-replication-6.5-1.el7.x86_64) >Restarted glusterd (systemctl restart glusterd) >Restarted the problematic Gluster volume >Reattached the problematic storage domain >Started the problematic storage domain > >Things work now. I can now run VMs and write data, copy virtual disks, >move >virtual disks to other storage domains, etc. > >I am very thankful that the storage domain is working again ! > >How can I safely perform upgrades on Gluster ! When will it be safe to >do >so ? > >Thank You Again For Your Help ! > >On Mon, Jun 22, 2020 at 10:58 AM C Williams >wrote: > >> Strahil, >> >> I have downgraded the target. The copy from the problematic volume to >the >> target is going on now. >> Once I have the data copied, I might downgrade the problematic >volume's >> Gluster to 6.5. >> At that point I might reattach the original ovirt domain and see if >it >> will work again. >> But the copy is going on right now. >> >> Thank You For Your Help ! >> >> On Mon, Jun 22, 2020 at 10:52 AM Strahil Nikolov > >> wrote: >> >>> You should ensure that in the storage domain tab, the old storage >is >>> not visible. >>> >>> I still wander why yoiu didn't try to downgrade first. >>> >>> Best Regards, >>> Strahil Nikolov >>> >>> На 22 юни 2020 г. 13:58:33 GMT+03:00, C Williams > >>> написа: >>> >Strahil, >>> > >>> >The GLCL3 storage domain was detached prior to attempting to add >the >>> >new >>> >storage domain. >>> > >>> >Should I also "Remove" it ? >>> > >>> >Thank You For Your Help ! >>> > >>> >-- Forwarded message - >>> >From: Strahil Nikolov >>> >Date: Mon, Jun 22, 2020 at 12:50 AM >>> >Subject: Re: [ovirt-users] Re: Fwd: Fwd: Issues with Gluster Domain >>> >To: C Williams >>> >Cc: users >>> > >>> > >>> >You can't add the new volume as it contains the same data (UUID) as >the >>> >old >>> >one , thus you need to detach the old one before adding the new one >- >>> >of >>> >course this means downtime for all VMs on that storage. >>> > >>> >As you see , downgrading is more simpler. For me v6.5 was working, >>> >while >>> >anything above (6.6+) was causing complete lockdown. Also v7.0 was >>> >working, but it's supported in oVirt 4.4. >>> > >>> >Best Regards, >>> >Strahil Nikolov >>> > >>> >На 22 юни 2020 г. 7:21:15 GMT+03:00, C Williams >>> > >>> >написа: >>> >>Another question >>> >> >>> >>What version could I downgrade to safely ? I am at 6.9 . >>> >> >>> >>Thank You For Your Help !! >>> >> >>> >>On Sun, Jun 21, 2020 at 11:38 PM Strahil Nikolov >>> >> >>> >>wrote: >>> >> >>> >>> You are definitely reading it wrong. >>> >>> 1. I didn't create a new storage domain ontop this new volume. >>> >>> 2. I used cli >>> >>> >>> >>> Something like this (in your case it should be 'replica 3'): >>> >>> gluster volume create newvol replica 3 arbiter 1 >>> >>ovirt1:/new/brick/path >>> >>> ovirt2:/new/brick/path ovirt3:/new/arbiter/brick/path >>> >>> gluster volume start newvol >>> >>> >>> >>> #Detach oldvol from ovirt >>> >>> >>> >>> mount -t glusterfs ovirt1:/oldvol /mnt/oldvol >>> >>> mount -t glusterfs ovirt1:/newvol /mnt/newvol >>> >>> cp -a /mnt/oldvol/* /mnt/newvol >>> >>> >>> >>> #Add only newvol as a storage domain in oVirt >>> >>> #Import VMs >>> >>> >>> >>> I still think that you should downgrade your gluster packages!!! >>> >>> >>> >>> Best Regards, >>> >>> Strahil Nikolov >>> >>> >>> >>> На 22 юни 2020 г. 0:43:46 GMT+03:00, C Williams >>> >> >>> >>> написа: >>> >>> >Strahil, >>> >>> > >>> >>> >It sounds like you used a "System Managed Volume" for the new >>> >>storage >>> >>> >domain,is that correct? >>> >>> > >>> >>> >Thank You For Your Help ! >>> >>> > >>> >>> >On Sun, Jun 21, 2020 at 5:40 PM C Williams >>> > >>>
[ovirt-users] Re: Fwd: Re: Fwd: Fwd: Issues with Gluster Domain
Strahil, I have downgraded the target. The copy from the problematic volume to the target is going on now. Once I have the data copied, I might downgrade the problematic volume's Gluster to 6.5. At that point I might reattach the original ovirt domain and see if it will work again. But the copy is going on right now. Thank You For Your Help ! On Mon, Jun 22, 2020 at 10:52 AM Strahil Nikolov wrote: > You should ensure that in the storage domain tab, the old storage is not > visible. > > I still wander why yoiu didn't try to downgrade first. > > Best Regards, > Strahil Nikolov > > На 22 юни 2020 г. 13:58:33 GMT+03:00, C Williams > написа: > >Strahil, > > > >The GLCL3 storage domain was detached prior to attempting to add the > >new > >storage domain. > > > >Should I also "Remove" it ? > > > >Thank You For Your Help ! > > > >-- Forwarded message - > >From: Strahil Nikolov > >Date: Mon, Jun 22, 2020 at 12:50 AM > >Subject: Re: [ovirt-users] Re: Fwd: Fwd: Issues with Gluster Domain > >To: C Williams > >Cc: users > > > > > >You can't add the new volume as it contains the same data (UUID) as the > >old > >one , thus you need to detach the old one before adding the new one - > >of > >course this means downtime for all VMs on that storage. > > > >As you see , downgrading is more simpler. For me v6.5 was working, > >while > >anything above (6.6+) was causing complete lockdown. Also v7.0 was > >working, but it's supported in oVirt 4.4. > > > >Best Regards, > >Strahil Nikolov > > > >На 22 юни 2020 г. 7:21:15 GMT+03:00, C Williams > > > >написа: > >>Another question > >> > >>What version could I downgrade to safely ? I am at 6.9 . > >> > >>Thank You For Your Help !! > >> > >>On Sun, Jun 21, 2020 at 11:38 PM Strahil Nikolov > >> > >>wrote: > >> > >>> You are definitely reading it wrong. > >>> 1. I didn't create a new storage domain ontop this new volume. > >>> 2. I used cli > >>> > >>> Something like this (in your case it should be 'replica 3'): > >>> gluster volume create newvol replica 3 arbiter 1 > >>ovirt1:/new/brick/path > >>> ovirt2:/new/brick/path ovirt3:/new/arbiter/brick/path > >>> gluster volume start newvol > >>> > >>> #Detach oldvol from ovirt > >>> > >>> mount -t glusterfs ovirt1:/oldvol /mnt/oldvol > >>> mount -t glusterfs ovirt1:/newvol /mnt/newvol > >>> cp -a /mnt/oldvol/* /mnt/newvol > >>> > >>> #Add only newvol as a storage domain in oVirt > >>> #Import VMs > >>> > >>> I still think that you should downgrade your gluster packages!!! > >>> > >>> Best Regards, > >>> Strahil Nikolov > >>> > >>> На 22 юни 2020 г. 0:43:46 GMT+03:00, C Williams > >> > >>> написа: > >>> >Strahil, > >>> > > >>> >It sounds like you used a "System Managed Volume" for the new > >>storage > >>> >domain,is that correct? > >>> > > >>> >Thank You For Your Help ! > >>> > > >>> >On Sun, Jun 21, 2020 at 5:40 PM C Williams > > > >>> >wrote: > >>> > > >>> >> Strahil, > >>> >> > >>> >> So you made another oVirt Storage Domain -- then copied the data > >>with > >>> >cp > >>> >> -a from the failed volume to the new volume. > >>> >> > >>> >> At the root of the volume there will be the old domain folder id > >>ex > >>> >> 5fe3ad3f-2d21-404c-832e-4dc7318ca10d > >>> >> in my case. Did that cause issues with making the new domain > >>since > >>> >it is > >>> >> the same folder id as the old one ? > >>> >> > >>> >> Thank You For Your Help ! > >>> >> > >>> >> On Sun, Jun 21, 2020 at 5:18 PM Strahil Nikolov > >>> > > >>> >> wrote: > >>> >> > >>> >>> In my situation I had only the ovirt nodes. > >>> >>> > >>> >>> На 21 юни 2020 г. 22:43:04 GMT+03:00, C Williams > >>> > > >>> >>> написа: > >>> >>> >Strahil, > >>> >>> > > >>> >>> >So should I make the target volume on 3 bricks which do not > >have > >>> >ovirt > >>> >>> >-- > >>> >>> >just gluster ? In other words (3) Centos 7 hosts ? > >>> >>> > > >>> >>> >Thank You For Your Help ! > >>> >>> > > >>> >>> >On Sun, Jun 21, 2020 at 3:08 PM Strahil Nikolov > >>> > > >>> >>> >wrote: > >>> >>> > > >>> >>> >> I created a fresh volume (which is not an ovirt sgorage > >>> >domain), > >>> >>> >set > >>> >>> >> the original storage domain in maintenance and detached > >>it. > >>> >>> >> Then I 'cp -a ' the data from the old to the new volume. > >>Next, > >>> >I > >>> >>> >just > >>> >>> >> added the new storage domain (the old one was a kind > >>of a > >>> >>> >> 'backup') - pointing to the new volume name. > >>> >>> >> > >>> >>> >> If you observe issues , I would recommend you to > >>downgrade > >>> >>> >> gluster packages one node at a time . Then you might be > >>able > >>> >to > >>> >>> >> restore your oVirt operations. > >>> >>> >> > >>> >>> >> Best Regards, > >>> >>> >> Strahil Nikolov > >>> >>> >> > >>> >>> >> На 21 юни 2020 г. 18:01:31 GMT+03:00, C Williams > >>> >>> > > >>> >>> >> написа: > >>> >>> >> >Strahil, > >>> >>> >> > > >>> >>> >> >Thanks for the follow up ! > >>> >>> >> > > >>> >>> >> >How did you copy the data to another volu
[ovirt-users] Re: Fwd: Re: Fwd: Fwd: Issues with Gluster Domain
You should ensure that in the storage domain tab, the old storage is not visible. I still wander why yoiu didn't try to downgrade first. Best Regards, Strahil Nikolov На 22 юни 2020 г. 13:58:33 GMT+03:00, C Williams написа: >Strahil, > >The GLCL3 storage domain was detached prior to attempting to add the >new >storage domain. > >Should I also "Remove" it ? > >Thank You For Your Help ! > >-- Forwarded message - >From: Strahil Nikolov >Date: Mon, Jun 22, 2020 at 12:50 AM >Subject: Re: [ovirt-users] Re: Fwd: Fwd: Issues with Gluster Domain >To: C Williams >Cc: users > > >You can't add the new volume as it contains the same data (UUID) as the >old >one , thus you need to detach the old one before adding the new one - >of >course this means downtime for all VMs on that storage. > >As you see , downgrading is more simpler. For me v6.5 was working, >while >anything above (6.6+) was causing complete lockdown. Also v7.0 was >working, but it's supported in oVirt 4.4. > >Best Regards, >Strahil Nikolov > >На 22 юни 2020 г. 7:21:15 GMT+03:00, C Williams > >написа: >>Another question >> >>What version could I downgrade to safely ? I am at 6.9 . >> >>Thank You For Your Help !! >> >>On Sun, Jun 21, 2020 at 11:38 PM Strahil Nikolov >> >>wrote: >> >>> You are definitely reading it wrong. >>> 1. I didn't create a new storage domain ontop this new volume. >>> 2. I used cli >>> >>> Something like this (in your case it should be 'replica 3'): >>> gluster volume create newvol replica 3 arbiter 1 >>ovirt1:/new/brick/path >>> ovirt2:/new/brick/path ovirt3:/new/arbiter/brick/path >>> gluster volume start newvol >>> >>> #Detach oldvol from ovirt >>> >>> mount -t glusterfs ovirt1:/oldvol /mnt/oldvol >>> mount -t glusterfs ovirt1:/newvol /mnt/newvol >>> cp -a /mnt/oldvol/* /mnt/newvol >>> >>> #Add only newvol as a storage domain in oVirt >>> #Import VMs >>> >>> I still think that you should downgrade your gluster packages!!! >>> >>> Best Regards, >>> Strahil Nikolov >>> >>> На 22 юни 2020 г. 0:43:46 GMT+03:00, C Williams >> >>> написа: >>> >Strahil, >>> > >>> >It sounds like you used a "System Managed Volume" for the new >>storage >>> >domain,is that correct? >>> > >>> >Thank You For Your Help ! >>> > >>> >On Sun, Jun 21, 2020 at 5:40 PM C Williams > >>> >wrote: >>> > >>> >> Strahil, >>> >> >>> >> So you made another oVirt Storage Domain -- then copied the data >>with >>> >cp >>> >> -a from the failed volume to the new volume. >>> >> >>> >> At the root of the volume there will be the old domain folder id >>ex >>> >> 5fe3ad3f-2d21-404c-832e-4dc7318ca10d >>> >> in my case. Did that cause issues with making the new domain >>since >>> >it is >>> >> the same folder id as the old one ? >>> >> >>> >> Thank You For Your Help ! >>> >> >>> >> On Sun, Jun 21, 2020 at 5:18 PM Strahil Nikolov >>> > >>> >> wrote: >>> >> >>> >>> In my situation I had only the ovirt nodes. >>> >>> >>> >>> На 21 юни 2020 г. 22:43:04 GMT+03:00, C Williams >>> > >>> >>> написа: >>> >>> >Strahil, >>> >>> > >>> >>> >So should I make the target volume on 3 bricks which do not >have >>> >ovirt >>> >>> >-- >>> >>> >just gluster ? In other words (3) Centos 7 hosts ? >>> >>> > >>> >>> >Thank You For Your Help ! >>> >>> > >>> >>> >On Sun, Jun 21, 2020 at 3:08 PM Strahil Nikolov >>> > >>> >>> >wrote: >>> >>> > >>> >>> >> I created a fresh volume (which is not an ovirt sgorage >>> >domain), >>> >>> >set >>> >>> >> the original storage domain in maintenance and detached >>it. >>> >>> >> Then I 'cp -a ' the data from the old to the new volume. >>Next, >>> >I >>> >>> >just >>> >>> >> added the new storage domain (the old one was a kind >>of a >>> >>> >> 'backup') - pointing to the new volume name. >>> >>> >> >>> >>> >> If you observe issues , I would recommend you to >>downgrade >>> >>> >> gluster packages one node at a time . Then you might be >>able >>> >to >>> >>> >> restore your oVirt operations. >>> >>> >> >>> >>> >> Best Regards, >>> >>> >> Strahil Nikolov >>> >>> >> >>> >>> >> На 21 юни 2020 г. 18:01:31 GMT+03:00, C Williams >>> >>> > >>> >>> >> написа: >>> >>> >> >Strahil, >>> >>> >> > >>> >>> >> >Thanks for the follow up ! >>> >>> >> > >>> >>> >> >How did you copy the data to another volume ? >>> >>> >> > >>> >>> >> >I have set up another storage domain GLCLNEW1 with a new >>volume >>> >>> >imgnew1 >>> >>> >> >. >>> >>> >> >How would you copy all of the data from the problematic >>domain >>> >GLCL3 >>> >>> >> >with >>> >>> >> >volume images3 to GLCLNEW1 and volume imgnew1 and preserve >>all >>> >the >>> >>> >VMs, >>> >>> >> >VM >>> >>> >> >disks, settings, etc. ? >>> >>> >> > >>> >>> >> >Remember all of the regular ovirt disk copy, disk move, VM >>> >export >>> >>> >> >tools >>> >>> >> >are failing and my VMs and disks are trapped on domain GLCL3 >>and >>> >>> >volume >>> >>> >> >images3 right now. >>> >>> >> > >>> >>> >> >Please let me know >>> >>> >> > >>> >>> >> >Thank You For Your Help ! >>> >>> >> >