[Xen-users] [XCP] Version 1.1 - should quiesced snapshots work?

[Errol Neal <eneal@xxxxxxxxxxxxxxxxx>]

Hi. Please note that I am performing most operations using XenCenter 6. 

When taking a snapshot and checking the quiesced  box, the snapshot takes 
forever, eventually fails and while it is running, I get the following errors 
in /var/log/messages on the XCP host:

Mar 24 15:54:45 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
RescanThread: scanning targets...
Mar 24 15:54:45 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
XenbusFindVbds: ignoring cdrom (VBD 5696)
Mar 24 15:54:45 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
__XenvbdScanTargets: target 0 -> device/vbd/768
Mar 24 15:54:45 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
__XenvbdScanTargets: target 1 -> device/vbd/832
Mar 24 11:54:45 xenserver-bnfnijur vbd.uevent[add](backend/vbd/79/832): wrote 
/xapi/79/hotplug/vbd/832/hotplug = 'online'
Mar 24 11:54:46 xenserver-bnfnijur fe: 10158 
(/opt/xensource/xha/ha_query_liveset) exitted with code 0
Mar 24 15:54:51 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
DeviceRelationsFdo: scanning targets...
Mar 24 15:54:51 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
XenbusFindVbds: ignoring cdrom (VBD 5696)
Mar 24 15:54:51 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
__XenvbdScanTargets: target 0 -> device/vbd/768
Mar 24 15:54:51 xenserver-bnfnijur HVM79[7822]:  XENVBD   Notice   
__XenvbdScanTargets: target 1 -> device/vbd/832
Mar 24 11:55:02 xenserver-bnfnijur tapdisk[10137]: ERROR: errno -28 at 
vhd_complete: 
/dev/VG_XenStorage-3ee08436-2a81-cdda-f62a-291e67b517f8/VHD-dc8e12e3-b74f-427a-a59e-d934c0d8c33a:
 op: 5, lsec: 6160384, secs: 8, nbytes: 4096, blk: 1504, blk_offset: 4294967295
Mar 24 11:55:02 xenserver-bnfnijur tapdisk[10137]: ERROR: errno -28 at 
vhd_complete: 
/dev/VG_XenStorage-3ee08436-2a81-cdda-f62a-291e67b517f8/VHD-dc8e12e3-b74f-427a-a59e-d934c0d8c33a:
 op: 2, lsec: 6162424, secs: 8, nbytes: 4096, blk: 1504, blk_offset: 4294967295
Mar 24 11:55:02 xenserver-bnfnijur tapdisk[10137]: ERROR: errno -28 at 
__tapdisk_vbd_complete_td_request: req 0: write 0x0008 secs to 0x005e07f8
Mar 24 11:55:03 xenserver-bnfnijur tapdisk[10137]: ERROR: errno -28 at 
vhd_complete: 
/dev/VG_XenStorage-3ee08436-2a81-cdda-f62a-291e67b517f8/VHD-dc8e12e3-b74f-427a-a59e-d934c0d8c33a:
 op: 5, lsec: 6160384, secs: 8, nbytes: 4096, blk: 1504, blk_offset: 4294967295
Mar 24 11:55:03 xenserver-bnfnijur tapdisk[10137]: ERROR: errno -28 at 
vhd_complete: 
/dev/VG_XenStorage-3ee08436-2a81-cdda-f62a-291e67b517f8/VHD-dc8e12e3-b74f-427a-a59e-d934c0d8c33a:
 op: 2, lsec: 6162424, secs: 8, nbytes: 4096, blk: 1504, blk_offset: 4294967295
[SNIP]..

Dmesg also reports i/o errors errors to the tap:

end_request: I/O error, dev tdf, sector 6162424
end_request: I/O error, dev tdf, sector 6162424

I have the Citrix Tools installed as well as the VSS provider for Citrix. 

The funky thing that I observe is that the snapshot gets added as a disk to the 
VM. I'm unable to remove it and my only option is to deactivate it. After 
deactivating it, I can do xe snapshot-list and no snapshots are listed,but if I 
view the SR, I can clearly see that it's still here shown as "0% on disk" If I 
do a xe vdi-list, I can see two disks with the same name-label though. For 
example:

[root@xenserver-bnfnijur ~]# xe vdi-list name-label=VS2025_Disk0
uuid ( RO)                : 47d8d6de-a61c-4e1c-a708-014d9d08aa80
          name-label ( RW): VS2025_Disk0
    name-description ( RW): Disk 0 for VS2025
             sr-uuid ( RO): 3ee08436-2a81-cdda-f62a-291e67b517f8
        virtual-size ( RO): 80002154496
            sharable ( RO): false
           read-only ( RO): false


uuid ( RO)                : dc8e12e3-b74f-427a-a59e-d934c0d8c33a
          name-label ( RW): VS2025_Disk0
    name-description ( RW): Disk 0 for VS2025
             sr-uuid ( RO): 3ee08436-2a81-cdda-f62a-291e67b517f8
        virtual-size ( RO): 80002154496
            sharable ( RO): false
           read-only ( RO): false


The only way I've been able to restore things to a normal state is to actually 
destroy the VM.

Any thoughts?



_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users