Cloud Cache Lock Issue RRS feed

  • Question

  • We are experiencing a strange FSLogix Cloud Cache issue with our RDS 2019 environment;

    OS: Windows Server 2019

    FSLogix: 2.9.7349.30108

    We are using 25 Remote Desktop Session hosts based on Windows Server 2019, sometimes it happens that a user can't log on to all our session hosts, the error is:

    Attempting to acquire exclusive lock 'merel.hagedoorn\Profile_merel.hagedoorn.VHDX.lock' (0 of 60000 ms timeout).
    [ERROR:00000000]           Failed to acquire exclusive lock 'merel.hagedoorn\Profile_merel.hagedoorn.VHDX.lock'. SMBLockStorage.cpp(58): [WCODE: 0x00000005] Unable to open/create the lock file: \\zorgnet\dfs\RDS-PROFILES\merel.hagedoorn\Profile_merel.hagedoorn.VHDX.lock

    The share of our FSProfiles is published trough DFS: \\zorgnet\dfs\RDS-PROFILES, the default share is: \\RDS-profiles\Profiles

    Our FSLogix Settings:

    In this case for example this user can logon  to RDS01:

    But the user can't logon to most other RDS's, then we get this in ProcMon (in this case RDS12):

    As you can see, it looks like the error is generated when the frxccds.exe process wants to use the DFS path instead of the direct path, but the user can write to this directory throught the DFS path and the direct path.

    Other users that are logon to the RDS that is generating this error arn't facing any issues and for that users the direct path is show in ProcMon instead of the DFS Path.

    The configuration of all of the RDS servers are indentical.

    Any help would be appreciated

    Tuesday, June 9, 2020 11:34 AM

All replies

  • Hi Danielson,

    we are also seeing this issue in our new environment we are currently building.

    Things I see which are different to your config:
    We additionally have the following settings:
    - ConcurrentUserSessions = 1 (which is mandatory for WS2019 afaik)
    - FlipFlopProfileDirectoryName = 1 (instead of custom folder naming)
    - ProfileType = 3 (support for read-only profiles we use in conjunction with Citrix Profile Management)
    - RoamSearch = 0 (as User-based Windows Search indexing is enabled in WS2019 and to prevent FSLogix handle this search index on top of that)
    - SetTempToLocalPath = 3
    - SizeInMBs = 10240 (10 GB)
    - For TestingPurposes I played around with: LockedRetryCount=2, LockedRetryInterval=3 but with no luck
    - No DFS fileservers, just two normal fileservers both named in the configuration

    Do you use DFS with multiple fileservers in the backend? If this is the case exclusive locks can't be handled afaik (how should one server know, if the other grants this lock?). But with Cloud Cache you can simply name two servers that will keep the same profile data (FSLogix will care for that).

    In some testing I did back in May I also captured a message in ProcMon:
    Class: Filesystem
    Operation: CreateFile
    Path: \\uncpath\to\profile.lock
    Desired Access: Generic Read/Write, Delete
    Options: Synchronous IO Non-Alert, ... (sorry did not capture all of this line)

    This happens during login attempts until I restart the VM. Then everything works fine again (so NTFS/file share permissions should be ok).

    Curiously this request was not visible on the file server itself. I would guess that it never left the VM towards the server. Do you also see this?

    The fact that everything is going to work again after a reboot of the VM leads me to look for the problem somewhere on the FSLogix RDS/Citrix machines itself.

    Anyone from Microsoft/FSlogix with more ideas? Any more information needed to get closer to a solution?

    At the moment this is still a show stopper for a production use case (at least for us) as it can only be solved by rebooting this multi user host.



    Tuesday, July 28, 2020 2:28 PM