Vertical backup 1.1.5 on ESXI 6.5, vertical runs out of memory running backup. Tried “–no-chunk-listing”, did not help.
|----Total Memory…5.00 MB
./vertical backup --no-chunk-listing fs
gchen
April 5, 2018, 2:25am
#2
How big is the mm-flat disk?
mm-flat.vmdk is approx 1.65TB
gchen
April 8, 2018, 1:37am
#4
I’ll release a new version next week that should use much less memory. I’ll let you know when it is ready.
gchen
April 14, 2018, 2:44am
#5
Sorry I hasn’t been able to work on the new version this week. Should be able to get it done early next week. I’ll send you a new 30-day trial license when the new version is out.
gchen
April 18, 2018, 3:11am
#6
I uploaded a new version 1.2.0 at http://acrosync.com/esxi/vertical .
This version should be much more memory efficient than previous versions.
If you log into your customer page you should also see a 30-day license there. You can activate it by running vertical license activate and then entering the license code.
Hi…
|----Total Memory......24.00 MB
|----Reserved Memory...2.00 MB
|----Touched...........1.12 MB
|----Total Minimum (base)1.22 MB
|----Total Maximum (base)1.22 MB
|----Effective Minimum.2.20 MB
|----Target Allocation.1.22 MB
|----Total Memory......773.00 MB
|----Reserved Memory...751.00 MB
|----COW...............27.33 MB
|----Mapped............720.88 MB
|----Touched...........68.03 MB
|----Total Minimum (base)750.18 MB
|----Total Maximum (base)750.18 MB
|----Effective Minimum.751.62 MB
|----Target Allocation.750.18 MB
[root@localhost:/vmfs/volumes/59600a57-09287b8c-d7b0-0025904656a9/verticalbackup] ./vertical backup fs
gchen
April 20, 2018, 2:17am
#8
You can try to increase the chunk size by initializing the storage this way:
./vertical init -c 4M hostid sftp://esxibackup@192.168.24.6/esxi
This should reduce the size of the chunk list in memory. The downside is that the deduplication will become less efficient with a larger chunk size. However, I think the 4MB chunk size is reasonable for disks larger than 1TB (this is the default average chunk size in Duplicacy).
Seems to be working better with the larger chunksize.
