This post is more than 5 years old
1 Rookie
•
47 Posts
0
52526
November 8th, 2015 12:00
Equallogic used space after rethinning differs in OS from EQL group view
HI,
i have a PS4100 storage backend that serves a 1TB thin volume LUN to a Windows 2008R2 server. The volume is formatted with 64Kb NTFS blocksize
In windows the Host Integration Toolkit is installed, and the rethinning option is scheduled regularly.
Still after rethinning, the OS displays 260GB free space on the volume, but the EQL group manager displays only 90GB free space.
There is a difference of 170GB between the two. Where is all the overhead going? Could the 64KB blocksize generate this much overhead into the backend? and how could i generate statistics or calculate the overhead? The number of files on the volume is 5548. So normal usage for this volume are large files. (on average 137Mb/file)
Can anyone explain a little further into this?
MoRRiS11
1 Rookie
•
47 Posts
0
November 10th, 2015 09:00
The volume is not replicated. Firmware level is 7.1.8
and the sectorsize of the volume on the equallogic is 512bytes.
To see the impact i just started a defragment job on the windows file system and want to do a rethin afterwards to see what the numbers will be.
Just trying to understand the numbers behind it :)
MoRRiS11
1 Rookie
•
47 Posts
0
November 10th, 2015 10:00
Yes,
from within the MS2008 server we have an iSCSI initiator directly connecting to a PS4100 where the LUN is residing.
The MS2008 server has the HIT kit installed.
MoRRiS11
1 Rookie
•
47 Posts
0
November 10th, 2015 14:00
Hi,
the HIT version is 4.7.1, and the host does have access to the VSS-control volume
in the meantime I also defragmented the disk.
S:\>eqlrethin -a s:
Volume s:
File system reports 765.1 GB in use, SAN reports 1.0 TB allocated for this vol
ume.
Rethinning could recover up to 0.0 MB (0%) of storage for this volume.
Rethinning is not recommended.
Right now the Dell GUI (Dell Equallogic Auto Snapshot Manager) reports 258.8GB free space in the OS,
and 227.83GB Free on the PS
So the difference in numbers is much smaller then before.
Still trying to understand the difference :)
MoRRiS11
1 Rookie
•
47 Posts
0
November 11th, 2015 06:00
ok, so i am starting to understand the numbers. Especially the 15MB page sizes are interesting. (The fact that there is a difference is not an issue, just trying to understand it :)
the filesystem has about 5000 files, so if fragmentation is not to excessive it could even be possible to have up to 5000x15MB is 75GB of overhead.
so this is more or less in line with what i see.
if there is a lot of fragmentation the numbers can go up much further :)
Tnx for explaining.