01.07.2016 09:23, Lindsay Mathieson ?????:> How did you do your downgrade?downloaded packages from https://download.gluster.org/pub/gluster/glusterfs/3.7/3.7.11/CentOS/epel-7.2/x86_64/ and https://download.gluster.org/pub/gluster/glusterfs/3.7/3.7.11/CentOS/epel-7.2/noarch/ and yum downgrade * from this directory> > Had you changed the op-version to 3071?It's strange, but we have operating-version=30710 in /var/lib/glusterd/glusterd.info so I decided to not change it. I can mount volumes using fuse, but can't use libvirt, strange but true. If I write wrong pool name in VM config , then I get failed to initialize gluster connection to server: 'localhost': No such file or directory But if I write right pool name, but any file name then I get failed to initialize gluster connection to server: 'localhost': Invalid argument Looks like upgrade to 3.7.12 changed something in volumes config and this change is not compatible with gfapi, but I don't see any changes. Here is volume info for volume with VM images: Volume Name: pool Type: Replicate Volume ID: 6748e47c-4a2e-4a80-a42d-61b6917f4a41 Status: Started Number of Bricks: 1 x 3 = 3 Transport-type: tcp Bricks: Brick1: father:/wall/pool/brick Brick2: son:/wall/pool/brick Brick3: spirit:/wall/pool/brick Options Reconfigured: network.ping-timeout: 10 performance.readdir-ahead: on performance.quick-read: off performance.read-ahead: off performance.io-cache: off performance.stat-prefetch: off network.remote-dio: enable cluster.quorum-type: auto cluster.server-quorum-type: server cluster.eager-lock: enable nfs.disable: on storage.owner-uid: 107 storage.owner-gid: 107 auth.allow: 127.0.0.1,192.168.22.26,192.168.22.27,192.168.22.28 features.barrier: disable May be there are any ideas what we can try to change?> > On 1 July 2016 at 15:03, Dmitry Melekhov <dm at belkam.com> wrote: >> 30.06.2016 15:42, Dmitry Melekhov ?????: >> >>> 30.06.2016 15:40, Lindsay Mathieson ?????: >>>> On 30/06/2016 6:54 PM, Dmitry Melekhov wrote: >>>>> After upgrade from 3.7.11 to 3.7.12 we can't start VMs after VM >>>>> shutdown: >>>>> >>>>> virsh create w8test.xml >>>>> error: Failed to create domain from w8test.xml >>>>> error: failed to initialize gluster connection to server: 'localhost': >>>>> Invalid argument >>>> >>>> This looks similar to the errors we are seeing in the "3.7.12 disaster" >>>> thread, with the debian 3.7.12 packages. For myself things were fine with a >>>> rolling ugrade and running VM's, until we shut them down. >>>> >>> Thank you! >>> >>> I guess we have the same issue here, although I see no errors in logs. >>> >>> As I see you reverted back to 3.7.11, we are planning to go this way >>> tomorrow :-( >>> >> Downgrade to 3.7.11 doesn't help in our case, don't know why yet. >> fuse mount works OK. But virsh still says error: failed to initialize >> gluster connection to server: 'localhost': Invalid argument. >> Looks like upgrade to 3.7.12 changed some lib, which was not upgraded >> before. >> >> >> _______________________________________________ >> Gluster-users mailing list >> Gluster-users at gluster.org >> http://www.gluster.org/mailman/listinfo/gluster-users > >
01.07.2016 09:46, Dmitry Melekhov ?????:> 01.07.2016 09:23, Lindsay Mathieson ?????: >> How did you do your downgrade? > > downloaded packages from > https://download.gluster.org/pub/gluster/glusterfs/3.7/3.7.11/CentOS/epel-7.2/x86_64/ > > and > https://download.gluster.org/pub/gluster/glusterfs/3.7/3.7.11/CentOS/epel-7.2/noarch/ > > > and > yum downgrade * > from this directoryOK, forgot to restart libvirtd after this downgrade. Rebooted one of servers to be completely sure. Now I get virsh create w8test.xml error: Failed to create domain from w8test.xml error: End of file while reading data: Input/output error :-( looks like some problems with libgfs packages for centos...
Lindsay Mathieson
2016-Jul-01 23:02 UTC
[Gluster-users] 3.7.12, centos 7 , libvirt problem
On 1/07/2016 3:46 PM, Dmitry Melekhov wrote:> It's strange, but we have > operating-version=30710 > in /var/lib/glusterd/glusterd.infoChiming in late, but the op-version has to be manually upgraded, it won't happen automatically. A safety precaution basically. -- Lindsay Mathieson