Re: [ClusterLabs] DRBD not failing over
Quoting Jaap Winius : Very interesting. I'm already running DRBD 9, so that base has already been covered, but here's some extra information: My test system actually consists of a single 4-node DRBD cluster that spans two data centers, with each data center having a 2-node Pacemaker cluster to fail resources over between the two DRBD nodes in that data center. But, for the purpose of quorum arbitration I guess these extra DRBD nodes don't matter, perhaps because four is not an odd number? No, four nodes are enough and I eventually figured it out. There was an extra firewall port that may have helped (2224), but I suspect that the main problem was that I forgot to enable the SELinux boolean for DRBD: daemons_enable_cluster_mode=1. Now everything is working perfectly. Thanks, Jaap ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] DRBD not failing over
On February 26, 2020 2:36:46 PM GMT+02:00, "Nickle, Richard" wrote: >I spent many, many hours tackling the two-node problem and I had >exactly >the same symptoms (only able to get the resource to move if I moved it >manually) until I did the following: > >* Switch to DRBD 9 (added LINBIT repo because DRBD 8 is the default in >the >Ubuntu repo) >* Build a third diskless quorum arbitration node. > >My DRBD configuration now looks like this: > >hatst2:$ sudo drbdadm status > >r0 role:*Primary* > > disk:*UpToDate* > > hatst1 role:Secondary > >peer-disk:UpToDate > > hatst4 role:Secondary > >peer-disk:Diskless > >On Wed, Feb 26, 2020 at 6:59 AM Jaap Winius wrote: > >> >> Hi folks, >> >> My 2-node test system has a DRBD resource that is configured as >follows: >> >> ~# pcs resource defaults resource-stickiness=100 ; \ >> pcs resource create drbd ocf:linbit:drbd drbd_resource=r0 \ >> op monitor interval=60s ; \ >> pcs resource master drbd master-max=1 master-node-max=1 \ >> clone-max=2 clone-node-max=1 notify=true >> >> The resource-stickiness setting is to prevent failbacks. I've got >that >> to work with NFS and and VIP resources, but not with DRBD. Moreover, >> when configured as shown above, the DRBD master does not even want to >> fail over when the node it started up on is shut down. >> >> Any idea what I'm missing or doing wrong? >> >> Thanks, >> >> Jaap >> >> PS -- I can only get it to fail over if I first move the DRBD >resource >> to the other node, which creates a "cli-prefer-drbd-master" location >> constraint for that node, but then it ignores the resource-stickiness >> setting and always performs the failbacks. >> >> PPS -- I'm using CentOS 7.7.1908, DRBD 9.10.0, Corosync 2.4.3, >> Pacemaker 1.1.20 and PCS 0.9.167. >> >> ___ >> Manage your subscription: >> https://lists.clusterlabs.org/mailman/listinfo/users >> >> ClusterLabs home: https://www.clusterlabs.org/ >> >> Is your DRBD used as LVM PV -> like as a disk for iSCSI LUN ? If yes, ensure that you have an LVM global filter for the /dev/drbdXYZ and the physical devices (like /dev/sdXYZ ) and the wwid . Best Regards, Strahil Nikolov ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] DRBD not failing over
Quoting "Nickle, Richard" : * Switch to DRBD 9 ... * Build a third diskless quorum arbitration node. Very interesting. I'm already running DRBD 9, so that base has already been covered, but here's some extra information: My test system actually consists of a single 4-node DRBD cluster that spans two data centers, with each data center having a 2-node Pacemaker cluster to fail resources over between the two DRBD nodes in that data center. But, for the purpose of quorum arbitration I guess these extra DRBD nodes don't matter, perhaps because four is not an odd number? Cheers, Jaap ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
Re: [ClusterLabs] DRBD not failing over
I spent many, many hours tackling the two-node problem and I had exactly the same symptoms (only able to get the resource to move if I moved it manually) until I did the following: * Switch to DRBD 9 (added LINBIT repo because DRBD 8 is the default in the Ubuntu repo) * Build a third diskless quorum arbitration node. My DRBD configuration now looks like this: hatst2:$ sudo drbdadm status r0 role:*Primary* disk:*UpToDate* hatst1 role:Secondary peer-disk:UpToDate hatst4 role:Secondary peer-disk:Diskless On Wed, Feb 26, 2020 at 6:59 AM Jaap Winius wrote: > > Hi folks, > > My 2-node test system has a DRBD resource that is configured as follows: > > ~# pcs resource defaults resource-stickiness=100 ; \ > pcs resource create drbd ocf:linbit:drbd drbd_resource=r0 \ > op monitor interval=60s ; \ > pcs resource master drbd master-max=1 master-node-max=1 \ > clone-max=2 clone-node-max=1 notify=true > > The resource-stickiness setting is to prevent failbacks. I've got that > to work with NFS and and VIP resources, but not with DRBD. Moreover, > when configured as shown above, the DRBD master does not even want to > fail over when the node it started up on is shut down. > > Any idea what I'm missing or doing wrong? > > Thanks, > > Jaap > > PS -- I can only get it to fail over if I first move the DRBD resource > to the other node, which creates a "cli-prefer-drbd-master" location > constraint for that node, but then it ignores the resource-stickiness > setting and always performs the failbacks. > > PPS -- I'm using CentOS 7.7.1908, DRBD 9.10.0, Corosync 2.4.3, > Pacemaker 1.1.20 and PCS 0.9.167. > > ___ > Manage your subscription: > https://lists.clusterlabs.org/mailman/listinfo/users > > ClusterLabs home: https://www.clusterlabs.org/ > > ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
[ClusterLabs] DRBD not failing over
Hi folks, My 2-node test system has a DRBD resource that is configured as follows: ~# pcs resource defaults resource-stickiness=100 ; \ pcs resource create drbd ocf:linbit:drbd drbd_resource=r0 \ op monitor interval=60s ; \ pcs resource master drbd master-max=1 master-node-max=1 \ clone-max=2 clone-node-max=1 notify=true The resource-stickiness setting is to prevent failbacks. I've got that to work with NFS and and VIP resources, but not with DRBD. Moreover, when configured as shown above, the DRBD master does not even want to fail over when the node it started up on is shut down. Any idea what I'm missing or doing wrong? Thanks, Jaap PS -- I can only get it to fail over if I first move the DRBD resource to the other node, which creates a "cli-prefer-drbd-master" location constraint for that node, but then it ignores the resource-stickiness setting and always performs the failbacks. PPS -- I'm using CentOS 7.7.1908, DRBD 9.10.0, Corosync 2.4.3, Pacemaker 1.1.20 and PCS 0.9.167. ___ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
