Continuous Replication - problem on coalesce
-
Hi,
I use XCP-NG 8.2, up to date.
XOCE - commit bc6afc3933b72e172902489b639c42a366cf4f13I configure continuous replication of a VM, when I make a full backup everything is ok (the size coincides with the original), when it starts the backup delta, everything is ok until coalesce starts, when it keeps only the last delta and deletes the base ( full backup ).
The result being a clone with only a difference, unfunctional.Ex:
VM has an HDD with a total size of 100gb, occupied 50gb,
for some operations I write 2gb, the result being 52gb.
Initially when I do full backup, on an ext4, it will make a 50gb vhd ( all is ok ),
after I have performed the operations I start a delta backup, the final result is an 2gb vhd.
The final result had to be 52gb not 2gb.The same thing happens on the LVM format,
and keep only the last delta backup, not a total backup with base + deltaThank you
-
Hi gheppy,
Have you tried starting the VM to see if you have an error that could help us?
-
I cloned the result and started the VM, it can't find the boot hdd.
I mounted the resulting disk in another VM with a good operating system, the result being a disk that is not formatted and not initialized. -
Hmm weird, have we broke anything on
masterregarding that @julien-f ?Anyone else could try to reproduce this behavior?
-
the problem start with commit 9139c5e9d6b4306ba4078e6fa128a36f65417792
-
Thanks, we are taking a look right now.
-
We try to spot a problem but we couldn't reproduce.
Can you describe the problem functionally speaking? (not from what's happening behind in your opinion)
-
I have the following configurations:
1.
An XCP-NG server, on which I have a VM with an application (a windows) and a VM with XOCE,
On the server I attached a NAS (synology) on iSCSI with LVM.
Every day there is a backup with CR and once a week a full backupAn XCP-NG server, on which I have a VM with an application (a windows)
and A server on which I have a VM with XOCE, connected to each other on the 10gb switch.
On the second server I attached a NAS (TrueNAS core) on iSCSI at operating system level and then mounted in XCP-ng as local ext3.
Every day there is a backup with CR and once a week a full backupThe behavior is difficult to detect on LVM, always having the same size.
This behavior can be seen on ext3 because there you can see vhd and its size.I will make some print screen with the configurations
-
XOCE is on Ubuntu 20.04 and is installed according to the documentation.
At the moment everything works as before with commit 65e62018e6aeabf64cf9def01ad043dfbdff5c18In log everything seems ok, only the physical result ( vhd ) is wrong
to understand the abbreviations:
RAID - is the local raid of the server
iSCSI - is the connection to the NAS
iSCSI-ext - is the connection to the NAS at the level of the XCP-NG operating system and mounted in XCP-NG as local-extfirst configuration
photo1
photo2
photo3
second configuration
photo1
photo2
photo3
photo4
photo5
-
But what behavior? It's still unclear to me what is the actual problem
Please describe the problem itself. You can't boot a cloned replicated VM?
-
No, I can not.
The result of a clone is a VM with two disks without content and partitions, one where the operating system should be and the second where the database should be.
I have attached the two disks to a VM with a functional operating system ( windows ) and both appear unformatted and uninitialized.
What two disks have as the occupied space on the XCP server's ext partition ( after clone ) are the only the value of the last CR has. -
If you try to boot directly the replicated VM (force boot it), does it work?
-
No, I will make a video with the problem for you to see
-
On the link below is the video with the problem.
Server is with all hdd on local controller, no iSCSI -
Thanks for the video, it's more clear now
And you don't get this issue at all with the commit before?
-
yes, I have no problem with commit before.
For now I'm on commit 65e62018e6aeabf64cf9def01ad043dfbdff5c18 and all is ok,
