Continual disconnects from AWS Broker ThinOS 9.1.2102

Tagged: ,

Viewing 15 posts - 1 through 15 (of 21 total)
  • Author
    Posts
  • #105498
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    Our users are continually getting disconnected from their broker session after the upgrade to 9.1.2102. They can connect for about 5 minutes then it just kicks them out to the ThinOS desktop where they have to log back in again.

    Event Log shows

    [PCoIP] disconnect “<remote pc ip address>

    [Amazon] connection to broker signoff “<my registration code>

    Anyone else seeing this?

    #105550
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    We are also experiencing disconnects randomly for users throughout the day. I’ve exported all logs from the terminal and have been going through them. I have not found anything helpful yet.

    [2021-06-09 07:09:50.111] N/WebUIProc/WebUIModule (1/1) [PCoIP] disconnect “***”
    [2021-06-09 07:10:19.367] N/WebUIProc/WebUIModule (1/1) [Amazon] connection broker signoff “***”
    [2021-06-09 07:10:20.826] N/WebUIProc/WebUIModule (1/1) [Amazon] login to broker “code://***”
    [2021-06-09 07:10:29.551] N/WebUIProc/WebUIModule (1/1) [PCoIP] start up “***”
    [2021-06-09 07:10:29.555] N/WebUIProc/WebUIModule (1/1) [PCoIP] connecting to “***”

    #105551
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    This might be the same as we are seeing.  Our clients don’t seem to actually disconnect from the broker, just the session – i.e. they are returned to the thinOS desktop and the PCoIP session/connection icon can still be seen on the desktop.

    One thing we have noticed is that the time zone on the client (the aws workspace) keeps getting changed to UTC after each disconnect.

    We think this has something to do with TimeZone redirection.  Our users are GMT Standard and the Dell box is too.  The physical machine (wyse 3040 in this case) is supposed to send the user’s time zone to the client which, in turn uses this to provide a localised pc experience.  This is part of Teradidci PCoIP drivers.  The time zone of UTC doesn’t make sense as you can’t live in UTC.

    There are instructions to disable this feature from Amazon but the problem still persists for us even after we’ve switched this off. Amazon are trying to help us understand what is happening however,  when our users connect to the aws workspace via any other mechanism (e.g. aws software client on laptop/ipad, thinOS 8.6 etc) they don’t have this problem which strongly points at a ThinOS 9 problem.

    Dell have been next to useless however, even telling me that they don’t have an Amazon account so they can’t test it PCoIP with AWS.  They’ve not responded to me for 6 days.

    Dell have really let us down here – not only is the software terrible but the support has been abysmal.

    We’ve had to go and buy every member of staff upgraded to ThinOS 9 a laptop and get them to use the software client.  This is obviously a short term fix but we’ll definitely be moving away from Dell as a result.

     

     

     

    #105553
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    I should add that we set time zone via GPO with a log on script which works as expected but then after a disconnect, the time zone is set back to UTC.

    #105568
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    Sorry for a delayed reply on this. I typed out a reply yesterday but lost it due to a random hardware issue 🙁

    I think you’re correct that this is potentially the same issue. I’m not sure why, but our folks are reporting that the disconnect issue seems to be mainly when they are using Microsoft Office apps; specifically, Word & Excel.

    I’ve gathered log files from the terminals and the Windows PCoIP agent that are having the disconnect issue but haven’t been able to find anything helpful 🙁

    I had a report this morning that one person got disconnected while trying to do a copy paste into Excel, so I’ve disabled PCoIP clipboard redirection via GPO in hopes that this might help the disconnect issue.

    We were having a similar issue with the time zone. I was able to solve this by group policy with setting the time zone (via registry key) and disabling time zone redirection (in the same group policy).

    Not sure if your logon script for updating the time zone sets the registry key, but if not perhaps give this a try.

    Here’s the page I used for setting the time zone via GPO with a registry key.

    https://sonjtinmar.wordpress.com/2018/02/07/set-time-zone-through-group-policy/

     

    #105569
    brian1020
    Participant
    • Total Post: 259
    • Jacked into The Matrix
    • ★★★★★★

    I don’t use PCoIP or AWS so I’m not much help there. I am curious, though, since you’re saying the logs aren’t helpful; are you getting the logs from WMS using Request Log File or exporting logs locally to USB thumb drive?  Typically the exported logs are more robust and could provide more information. You can also have any of the impacted associates run a network trace from the thin client and reproduce the issue and it will save locally to the thin client until you export log files from that thin client.

    Anything stick out in the Windows event logs at the time of the disconnects?

    Can Amazon provide any network packet captures from their side?

    #105571
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    I’ve got the logs from USB from the Wyse terminal and from the “local” AWS Windows installation. There was nothing notable in the logs around the timeframe when the disconnect happened that I noticed, but there are a lot of different log files. Is there a particular log file I should be looking at that might be helpful? I recently bumped up the verbosity on both the logs (via WMS and via GPO) so I’m hoping to potentially get some additional information from that.

    #105576
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    We also see that keyboard/mouse interaction (including clipboard) seems to trigger the disconnect however, I think this is secondary to the initial problem which is the timezone.

    I think that once the timezone is wrong, the whole session gets very confused and there’s also some issues with re-connecting.

    Just to re-iterate, these problems do not occur when a user connects via a different client – e.g. aws software client on a laptop or even an iPad.

     

    #105577
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    Also, we have also tried setting the time zone directly in the Registry via GPO too but this had mixed results and the switch back to UTC always happened anyway.

    FYI – This is the process we followed to disable time zone redirection from Amazon but it doesn’t seem to work – https://docs.aws.amazon.com/workspaces/latest/adminguide/group_policy.html#gp_time_zone

     

    #105691
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    After much gathering of logs and back and forth with AWS support, they determined that the issue is a crash from Teradici PCoIP client. Details below.

     

    The reference log lines:
    * Client logs (compat\linux\apps\PCoIP\var\log\client) show a weird cursor error and just stop logging until the user reconnects:
    0121-06-16T09:42:00> gtk_mouse_btn: down GDK btn 1, Send state 1
    0121-06-16T09:42:00> gtk_mouse_btn: up GDK btn 1, Send state 0
    0121-06-16T09:42:01> gtk_mouse_btn: down GDK btn 1, Send state 1
    0121-06-16T09:42:01> gtk_mouse_btn: up GDK btn 1, Send state 0
    0121-06-16T09:42:12> gtk: DID NOT CREATE pixbuf from the shape, using software cursor.
    0121-06-16T09:04:46> pu_msg_init: 9943

    * Windows Application log generates 88 and 89 events for PCoIPAgentService with the following messages:
    The system cannot start another process at this time.
    A write fault occurred on the network.

    Since the workspace disconnects, it implicates client-side issue which must be fixed by the team that owns the Teradici PCoIP connection manager (which is Teradici). They do have newer versions of the software[1] and that would be my suggestion for you to update the software on your Client.

    Before proceeding further I would request you to understand that installation and configuration of third party product is out of scope for AWS Premium Support as we do not have visibility of third party product architecture and requirement.
    Please refer:- https://aws.amazon.com/premiumsupport/faqs/

    The zero client is owned by another company (Teradici) and I have enough information to conclude there is an issue with the client in this case (even without confirming it’s a cursor problem).

     


    @ConfGen
    Any idea when Dell will update Wyse PCoIP firmware? AWS support has told us  Perhaps the version numbers do not align? If they do, it appears Dell is still using a very outdated version of Teradici software. Teradici is at version 21.01.6 but the latest at Dell support for PCoIP is 20.07.6.25. We are really struggling here with this issue. I appreciate anything you can do on this one.

     

    https://www.dell.com/support/home/en-us/product-support/product/wyse-3040-thin-client/drivers

    https://docs.teradici.com/documentation/connection-manager-and-security-gateway/release-notes?showRelease=21.01.3

     

    #105693
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    Good work getting to this point and thanks for updating here.

    I have an open support case with Dell and I’ll mention this to them too and see if I can get them to update the driver.

    Unfortunately it’s painfully difficult to get any response from them so I wouldn’t hold your breath.

    #105694
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    FYI,

    I did go down the route of looking at the Teradici drivers a while ago when I first had this issue but you need to have a support contract to be able to download the required drivers and even if you did manage to get hold of a copy from somewhere else it’s not possible to create your own package file to use with WMS (i asked that here a while ago https://technicalhelp.de/topic/update-pcoip-package)

    #105695
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    Sorry for any confusion. This ticket is with AWS support, not Dell. We have a support contract with AWS so I went there first. I haven’t opened a support case yet with Dell because I suspect as you said, it will be very hard to get responses from Dell. I am going to open one. I’ll report back anything helpful.

    #105800
    alex01
    Participant
    • Total Post: 38
    • Frequent Flyer
    • ★★★

    After a very, very long time and an unquantifiable number of chase emails to Dell I finally managed to get hold of an updated firmware and PCoIP driver – see link below

    https://dell.app.box.com/s/nmxv9urdf97lxaw55n4qmymkuubfgupf

    If you’re using WMS you’ll need OS_Web_9.1Full_v9.1.3112.zip and Teradici_PCoIP_21.03.1_9.pkg from the pkgs folder.  You’ll need to pull the firmware file out of the zip but the PCoIP package is the actual file you need.

    I can’t guarantee how long this link will be available as I’m not in control of it.

    The install is REALLY slow and a number of times it seems like it has hung (there’s just a black screen for ages) but just leave it alone and eventually it gets there.

    On the face of it it seems to have solved the issue for us but we continue to test.

    #105802
    Nate Pinchot
    Participant
    • Total Post: 10
    • Regular Joe
    • ★★

    Thanks so much for continuing to follow up on this, the information, and the link for the updates.

    Glad to hear the updates seem to have resolved the issues for you 🙂

    Got a message back from Dell that they have published an update for the Wyse 3040 terminal firmware as well as an updated Teradici PCoIP package.

    https://www.dell.com/support/home/en-us/product-support/product/wyse-3040-thin-client/drivers

    We’ve just deployed this update via WMS. We will monitor and report back. Fingers crossed this fixes the issue for us!

Viewing 15 posts - 1 through 15 (of 21 total)
  • You must be logged in to reply to this topic.