https://worldofeuc.com logo
Join Slack
Powered by
This message was deleted.
# microsoft-fslogix
s
This message was deleted.
t
Occurs on multiple servers, 6 times in a row. Nothing before it in the logs for 5 minutes.
l
Still seeing this Tim?
t
Yes. It continues to regularly occur. Over 35 VMs I am seeing about 150ish messages an hour.
l
:-( when did it start happening?
t
Not sure. We just started logging everything to a central server, and noticed it when going through our other FSL problems. Figured we should tackle one problem at a time 😃
s
I dont want to hijack but one simple question and then return to normal broadcast. When you guys are using cloud cache are you forced to store the entire container on the VDA?
t
No. Only cache of writes / reads.
l
No performance issues right now?
t
Not the ones we were having last week. Still having some of our servers (1 or 2) that will accept a logon connection and mount the VHDX. The user will not fully login, then they will log on to a second server, but their VHDX is still locked on the first, so they can't. This will continue ad nauseum until we reboot the first server. Even drainstopping doesn't work. We are also seeing Outlook profile problems (not using CC) - Outlook start in safe mode, missing emails, locking up, etc.
l
What are your file servers? Cluster or SOFS?
t
Neither. On the profile side just using CC to two separate 2016 core servers. On the ODFC side, single 2016 Core server.
l
Man.... any logs from the login process that sticks? Do you know where it's getting stuck?
t
Found this in the FSL event logs: Failed to read lock file: SMBStorage.cpp(809): [WCODE: 0x00000005] Unable to open/create the lock file: <file://DFSRoot/profiles/CPM/%3cusername%3e_%3cuser_SID%3e/Profile_%3cusername%3e.VHDX.lock%7C\\DFSRoot\profiles\CPM\&lt;username&gt;_&lt;user_SID&gt;\Profile_&lt;username&gt;.VHDX.lock>   Lock file read retry count exceeded   Failed to acquire an exclusive lock '<file://DFSROOT/profiles/CPM/%3cusername%3e_%3cuser_SID%3e/Profile_%3cusername%3e.VHDX.lock%7C\\DFSROOT\profiles\CPM\&lt;username&gt;_&lt;user_SID&gt;\Profile_&lt;username&gt;.VHDX.lock>': SMBStorage.cpp(809): [WCODE: 0x00000005] Unable to open/create the lock file: <file://DFSROOT/profiles/CPM/%3cusername%3e_%3cuser_SID%3e/Profile_%3cusername%3e.VHDX.lock%7C\\DFSROOT\profiles\CPM\&lt;username&gt;_&lt;user_SID&gt;\Profile_&lt;username&gt;.VHDX.lock>
The first error repeats a ton of times, until the lock retry is exceeded, then the bottom error is thrown.
l
Can you reproduce this? Have you done anything with AV? I'd double check exclusions and make sure VHDX files are in there. Server and client side.
t
No AV in the mix at all. None on the VDA, only Defender on the Core boxes, and it's disabled.
We can't intentionally reproduce, although it happens every day, on a different box
l
Backup?
t
I apologize Leee....not sure what you're asking. I don't believe it's tied to backups. We do have a backup of our PVS vDisk, although we are debating rolling a new one with older FSL version per @Eric.
l
No I mean, are there backups on the file servers. Sorry.
t
lol. Umm...Maybe? Checking...just going through this with another server that was just built...
l
Okay. It's the only other thing that could be locking.
t
Yup, no backups. Going to rectify that...
We moved our cache and write cache directories to the write cache persistent disk last night, and are still seeing these out of order warnings in the logs this morning.
l
So strange. Process monitor on your file server then to see the process locking the file I guess.
t
I think it's on the VDA side...but definitely could be on the file server side. Like I just tweeted you - I'm guessing next step is CAFS cluster.
l
Well....., what happens if you take a one side of the cloud cache out or revert to VHD locations and just use one of them for now?
t
Not sure. We discussed doing that, and it's easy enough. I'll see about changing it during our window tonight.
We did move the caches off to write cache disk on Tuesday night, and think that has helped a little.
l
Its cloud cache that's very hard to predict, it needs alot more work in my opinion. I remember attending the unveiling at a partner only meeting in the UK and I just had so many questions.
t
Odd thing is - it was stable with CC for several months. That was on the "bad" version (2.9.7205.27375). Search was broken with OneNote and Outlook together, though. So we updated, and think that may have been the proverbial straw.
Understood. I really want to like it. Thinking your idea of CAFS is the way to go.
l
I swear by it, either CAFS or SOFS with continuous availability. The only deciding factor is the underlying physical storage.
t
We've got two SANs, VMware on top. Should be fairly straight forward...
Downgraded FSL to 2.9.7117.<whatever> and will be testing today. Likely rolling to Prod tonight because of the gpresults problem in the other thread here.
Prod looks good today - still seeing the warning from the logs, yet no locked sessions or corrupt profiles or rebooted servers.
l
How strange Tim.
Open in Slack
PreviousNext
Powered by