Hello,
I recently ran into the "metadata too large for circular buffer problem" -> http://forum.proxmox.com/threads/12045-HELP!-metadata-too-large-for-circular-buffer
Because I have to keep the downtime minimal, what I want to do now is add an additional PV with a larger metadatasize to my shared VG, pvmove everything on the new PV, vgreduce and then pvremove the old PV. That should all be done while the VMs keep running.
Adding a PV to the VG and pvmoving works on-the-fly on a running proxmox cluster, I've already done that before. What I don't know is if the vgreduce/pvremove part may harm the shared LVM storage? Has anybody done anything like that before?
I recently ran into the "metadata too large for circular buffer problem" -> http://forum.proxmox.com/threads/12045-HELP!-metadata-too-large-for-circular-buffer
Because I have to keep the downtime minimal, what I want to do now is add an additional PV with a larger metadatasize to my shared VG, pvmove everything on the new PV, vgreduce and then pvremove the old PV. That should all be done while the VMs keep running.
Adding a PV to the VG and pvmoving works on-the-fly on a running proxmox cluster, I've already done that before. What I don't know is if the vgreduce/pvremove part may harm the shared LVM storage? Has anybody done anything like that before?