StarWind Software

Hello,

I have been testing the Starwind 8 Beta for a vSphere lab and noticed yesterday things were a bit slower than usual (been testing for about a week). Everything has been performing really well but yesterday I noticed in vSphere that latency on the LUN went up to about 5000 ms. The average according to the graph is about 2-5 ms. I only had about 5 active VM's running on this LUN and they dont really do much as far as disk access. I shut down the VM's on this particular datastore and rebooted the Starwind server. After reboot I opened the Starwind console and connected to the Starwind service which indicated that it was mounting the lsfs device. After about 15-20 minutes I received an errror "Device Mount Failed!". I tried rebooting again, still no luck. Looks to me like my VM's that were running on this datastore are history... Anyway to force mount the lsfs disk?

On another note, when I initially setup this test environment I created a lsfs disk and an imagefile disk to test performance. The imagefile disk never had an issue and mounts fine. Unfortunately I had my VM's running on the LSFS disk at the time. I also checked my RAID controller and it shows all disks healthy and RAID status Operational. My Windows server running Starwind never showed more than 2 GB's of memory utilized.

Info about my Starwind setup:

Windows Server 2012 R2
Starwind 8 Beta 3
Raid controller: Perc 6i
RAID: (4) disk RAID 10 Write Back enabled
Server has 4Gb RAM
CPU: Xeon L5639
Network: (3) 1Gbps NIC's (1 NIC is management, 2 NIC's for iSCSI)
iSCSI Setup: A side and b side on different VLAN's /seperate subnets - also modified the "<iScsiDiscoveryListInterfaces value="1"/" per the MPIO document
VMware showed both targets in the LUN path and I configured Round Robin for the Path Selection

LSFS Device Info:

Size 300GB
Cache enable: 256MB
Dedupe: No

Thanks in advance, look forward to your response.

Beta 3 is outdated. Please upgrade to Release Candidate as it would be quite stupid to hunt for bugs being (at least potentially) fixed in ~500 intermediate builds Thank you!

I had this problem too, but I have not been to reproduce it, leading me to a system stability issue and not a Starwind issue, my image files are fine, my LSFS files failed, it makes me wonder now as I write this if a “Reset button” during a write would produce the same results

Setup:

16GB Memory, 2 CPUs 2 Cores, 2008R2
V8 RC 8.0.6441
5 LUNs(in test) all sitting on same NTFS partition on SW iSCSI host, 128 WB cache no SSD cache, 10GB
1 IMG formatted to NTFS
1 LSFS with DeDup formatted to NTFS
1 LSFS without DeDup formatted to NTFS
1 LSFS with DeDup formatted to REFS
1 LSFS without DeDup formatted to NTFS

Test: ISO File Copies
Purpose Observe behavior of resource utilization, and active LSFS thin with delete

Copied 4 ISO files, (Windows Server 2012R2, 2008R2, Forefront TMG, Forefront UAG)
Repeated steps various times, all tests seems to only use 1 CPU but both cores will run up to 100%
1) Copied all 4(ISO) to the different LUNs one at a time, then Deleted (too big for cache)
2) Copied all 4(ISO) to all test LUNs at the same time, then deleted (too big for cache)
3) Copied 1 3GB ISO to LUNs one at a time, then deleted (all in cache)
4) Copied 1 3GB ISO to LUNs all at the same time, CRASH (really good throughput but still only 2 cores/1 CPU in use)

The CRASH turned into a reboot
All 4 LSFS failed to mount, all images were fine include the one in the test set
Removing and reattaching does not resolve the failed drives

Sounds like the exact same issue I have. I never did figure out a way to remount or recover the LSFS disk. I also tried upgrading to the RC and remount but it wont even recognize the files. I agree the throughput was pretty good using LSFS, but it definitely needs some form of recovery. I'm going to hang on to the LSFS files and see if it can be mounted in a later release but I doubt it. I noticed others on the forum mentioning extremely long mount times for small LUNs, but at least they mounted...

k0nf1gt wrote:Sounds like the exact same issue I have. I never did figure out a way to remount or recover the LSFS disk. I also tried upgrading to the RC and remount but it wont even recognize the files. I agree the throughput was pretty good using LSFS, but it definitely needs some form of recovery. I'm going to hang on to the LSFS files and see if it can be mounted in a later release but I doubt it. I noticed others on the forum mentioning extremely long mount times for small LUNs, but at least they mounted...

Please check this with RC build and let us know about the results.
Beta3 can not be updated to RC due to LSFS format changes.

I have uninstalled beta 3 and am now running RC. Will continue testing and will update if I can recreate the issue.

Hi!
May I ask if you have any new results to share with us?

On the RC everythings been stable so far although I havent been able to recover the old LSFS data. I also have only been using the imagefile devices...

k0nf1gt wrote:On the RC everythings been stable so far although I havent been able to recover the old LSFS data. I also have only been using the imagefile devices...

Thank you for keeping us updated.