I'm currently testing a Proxmox 2.1 HA setup with LVM on top of multipath iscsi devices (Equallogic SAN).
The snapshot backup job cannot remove the created lvm snapshot volume and so new job will fail, because of the already existing snapshot (interessing fact that the first backup final status will report as ok, altough he couldn't remove the snapshot).
I also tried to remove the snapshot by hand:
but this causes the IO to infinitely hang, which I found out was the fault of the multipath features "1 queue_if_no_path" option. So all the paths fail and cannot be rebuild, because the devices are still listed as busy.
After I removed this option the command immediately fails with following output:
If I connect to the same iscsi target on some other machine without multipath, I'm able to remove the volumes. Therefore I guess the error lies within multipath.
So my question is ... does anybody have hints how to deal with this error? Are there any known bugreports (couldn't find any)? Or is there anything in my configuration which is faulty?
Thanks.
The snapshot backup job cannot remove the created lvm snapshot volume and so new job will fail, because of the already existing snapshot (interessing fact that the first backup final status will report as ok, altough he couldn't remove the snapshot).
I also tried to remove the snapshot by hand:
Code:
lvremove /dev/equallogic/vzsnap-st00001-0but this causes the IO to infinitely hang, which I found out was the fault of the multipath features "1 queue_if_no_path" option. So all the paths fail and cannot be rebuild, because the devices are still listed as busy.
Code:
# multipath -v4
May 22 10:56:16 | libdevmapper: ioctl/libdm-iface.c(1705): device-mapper: reload ioctl on mpath0 failed: Device or resource busy
May 22 10:56:16 | libdevmapper: ioctl/libdm-iface.c(1705): device-mapper: reload ioctl on mpath0 failed: Device or resource busyAfter I removed this option the command immediately fails with following output:
Code:
# lvremove /dev/equallogic/vzsnap-st00001-0
/dev/disk/by-id/scsi-mpath1: BLKDISCARD ioctl at offset 61242081280 size 10737 41824 failed: Input/output error.
/dev/disk/by-id/scsi-mpath1: read failed after 0 of 4096 at 4096: Input/output errorIf I connect to the same iscsi target on some other machine without multipath, I'm able to remove the volumes. Therefore I guess the error lies within multipath.
So my question is ... does anybody have hints how to deal with this error? Are there any known bugreports (couldn't find any)? Or is there anything in my configuration which is faulty?
Thanks.
Last edited:
