StarWind Software

Bohdan (staff) wrote:Please try once again.

Now YES!!! version 8.0.6180

Thanks,

Got the new version, it now gives indication that an Image file failed

hmm... the LSFS LUNs failed to mount also
here is the new log file

Ironwolf,

1. ImageFile issues: ImageFile devices don't support disks with 4k block size now. Previous versions allowed to create devices on such disks, but devices could function incorrectly.
Now we have added check for 4k block size at the stage of device initialization.

2. one of structures of LSFS devices has been changed in the latest build of 8.0 beta. It led to incompatibility with LSFS devices that have been created with previous builds.
LSFS devices on the disk with 4k block size should function properly. You can just re-create LSFS devices.
It's possible to add backward-compatibility for this case, but we decided not to do it, as these are beta-builds, not releases.

Thank you for feedback!

barrysmoke wrote:I think I found the bottleneck. It's either a previously known passthrough issue with the lsi 9271-8i, and vmware 5.1, or it is an issue with firmware on my intel oem chassis. those are the only 2 things that are different, and I'm getting much better results testing on my second chassis.

I'll update here when I find out for sure.

confession...I found that it was my lack of exact replication of my testing environment. while previously testing stormagic, I had my iometer vm on a seperate ssd, and added a 100gig virtual disk on the iscsi datastore to be tested. when I moved on to test starwind, I had actually put the starwind vm on an ssd mirror, and I put my iometer vm on the same ssd mirror...looks like that was a no no. when I put the iometer vm back on a seperate ssd, I'm getting the fast results I expect, on par with my previous tests.

still having issue with ramdisk, and l1 cache of any kind, while starwind is running on a virtual machine inside a vmware 5.1 server.
i tried creating a 128gig ramdisk inside a vm, so I could compare my starwind san performance with a ramdisk i use a 100gig thick zero test disk added to my iometer vm for iops testing.....no go. vmware errors out when trying to format the new ramdisk datastore, invalid path.

I was able to create a 128Meg ramdisk, but when benchmarking, it kills the write performance. 100 iops range, instead of 19k to 50k depending on the disks I use.
in know this is an unusual request, but it would help to be able to benchmark starwind inside a vmware vm. I wonder if there is something I can do on the starwind vm configuration besides the memory reservation.
I did not deploy with the vsan option on starwind software. is there something that process does differently than just manually installing starwind v8 on a windows server vm?

here is what the log looks like:
http://bsmokeman.no-ip.biz/fileshare/12 ... 53.log.zip

should I start a new thread with the ramdisk problem under a vmware vm, or can you guys replicate it?

barrysmoke wrote:still having issue with ramdisk, and l1 cache of any kind, while starwind is running on a virtual machine inside a vmware 5.1 server.
i tried creating a 128gig ramdisk inside a vm, so I could compare my starwind san performance with a ramdisk i use a 100gig thick zero test disk added to my iometer vm for iops testing.....no go. vmware errors out when trying to format the new ramdisk datastore, invalid path.

I was able to create a 128Meg ramdisk, but when benchmarking, it kills the write performance. 100 iops range, instead of 19k to 50k depending on the disks I use.
in know this is an unusual request, but it would help to be able to benchmark starwind inside a vmware vm. I wonder if there is something I can do on the starwind vm configuration besides the memory reservation.
I did not deploy with the vsan option on starwind software. is there something that process does differently than just manually installing starwind v8 on a windows server vm?

here is what the log looks like:
http://bsmokeman.no-ip.biz/fileshare/12 ... 53.log.zip

Thanks for the feedback! We are checking it in our testlab and I believe we'll find the way to fix it.
As for the vSAN, currently there is now difference. You may deploy it manually installing StarWind inside the VM.

also, catch my thread on ramdisk, more details:
http://www.starwindsoftware.com/forums/ ... t3358.html

Sure. We do follow.

barrysmoke wrote:also, catch my thread on ramdisk, more details:
http://www.starwindsoftware.com/forums/ ... t3358.html

I made a comment about v8 on a 2008 server not having ramdisk functionality...I had tested between builds, and it was moved in the interface. my bad!
it's there, and doing another test.

Nice to hear you're fine now. Please keep us updated about your progress. Thank you!

barrysmoke wrote:I made a comment about v8 on a 2008 server not having ramdisk functionality...I had tested between builds, and it was moved in the interface. my bad!
it's there, and doing another test.

the ramdisk issue with errors, and dataloss was due to vaai. bohdan from starwind helped me disable vaai until a new build.
search gooogle for vaai, and you'll find the vmware page with all info, including how to disable.

Fixed bug related to VAAI so you'll have new build to test pretty soon. Thank you!

barrysmoke wrote:the ramdisk issue with errors, and dataloss was due to vaai. bohdan from starwind helped me disable vaai until a new build.
search gooogle for vaai, and you'll find the vmware page with all info, including how to disable.

Actually it is our additional detailed tracing and logging for VAAI in beta build. And, yes, it is already turned off (fixed).
I'll update the build soon.