Dedupe .spspx keeps growing

Public beta (bugs, reports, suggestions, features and requests)

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

jcrowe
Posts: 22
Joined: Wed Jan 08, 2014 1:49 am

Fri Mar 14, 2014 9:36 pm

I am now at 500GB used, while only 250 GB of the datastore shows used. I will be switching to thick disk until this gets sorted out, as I do not like the idea of having to create new deduped drives and migrating ~250GBs of data every few weeks.
User avatar
Anatoly (staff)
Staff
Posts: 1675
Joined: Tue Mar 01, 2011 8:28 am
Contact:

Thu Mar 20, 2014 11:03 am

Just a heads up for community: currently we are investigation the issue and fixing it.
We`ll keep you guys posted.
Best regards,
Anatoly Vilchinsky
Global Engineering and Support Manager
www.starwind.com
av@starwind.com
jcrowe
Posts: 22
Joined: Wed Jan 08, 2014 1:49 am

Thu May 08, 2014 9:01 pm

Has this been resolved with the new RC? I have been running with thick provisioning while waiting for this to be resolved. I would love to put LSFS under load with a couple database servers, without worrying that I will have to migrate it to a new drive every few weeks or so.
User avatar
anton (staff)
Site Admin
Posts: 4008
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Thu May 08, 2014 9:27 pm

Yes, RC2 has a fully functional garbage collector + TRIM and WRITE_SAME support. Please give it a try in Test & Development config to see would it work for you or not. Feedback is welcomed. Thank you!
jcrowe wrote:Has this been resolved with the new RC? I have been running with thick provisioning while waiting for this to be resolved. I would love to put LSFS under load with a couple database servers, without worrying that I will have to migrate it to a new drive every few weeks or so.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
jcrowe
Posts: 22
Joined: Wed Jan 08, 2014 1:49 am

Sat May 10, 2014 3:07 pm

anton (staff) wrote:Yes, RC2 has a fully functional garbage collector + TRIM and WRITE_SAME support. Please give it a try in Test & Development config to see would it work for you or not. Feedback is welcomed. Thank you!
jcrowe wrote:Has this been resolved with the new RC? I have been running with thick provisioning while waiting for this to be resolved. I would love to put LSFS under load with a couple database servers, without worrying that I will have to migrate it to a new drive every few weeks or so.
Here is my test setup and results:
I set up a HA lsfs 300GB drive, and moved 3 fairly high IO VM's on to it. After the VM's were moved to the drives, it showed 50GB used in VMWare, and 7 .spspx files were created (for a total of around 21GB's actually used). The Starwinds console showed that I was averaging about 50 iops per HA member. At this rate, it created a new .spspx file about every hour, and did not clean up any old ones. It is now up to 19 files created, which puts actual space used higher than space used if it was a thick drive.

Is garbage collection a scheduled task, and if so, can the frequency be changed? This is also probably not a very good gauge of a real world scenario, as I have only a few high IO VM's on one drive. Will this smooth out if other lower IO VM's are also hosted on this drive?
jcrowe
Posts: 22
Joined: Wed Jan 08, 2014 1:49 am

Sun May 11, 2014 5:46 pm

As an update to this, I moved the first VM that I moved to the lsfs drive back off the drive. The first VM was responsible for the first 3 or so .spspx files, so once it was removed from the drive, I would have assumed that those files would be released. It's now been 24 hours and not only are those files not removed, but new files keep being added.
User avatar
Bohdan (staff)
Staff
Posts: 435
Joined: Wed May 23, 2007 12:58 pm

Thu May 15, 2014 1:05 pm

Developers are working on this. I'll let you know as soon as the improvements are available.
Post Reply