Open Bug 1613357 Opened 5 years ago Updated 2 years ago

creates new profile after update. terminal services related?

Categories

(Toolkit :: Startup and Profile System, defect, P3)

71 Branch
Unspecified
Windows
defect

Tracking

()

UNCONFIRMED

People

(Reporter: manuel.rodrigues.gomes, Unassigned)

References

Details

Attachments

(4 files)

+++ This bug was initially created as a clone of Bug #1601287 +++
sorry for not giving any further information about the original bug report.

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:71.0) Gecko/20100101 Firefox/71.0

Steps to reproduce:

hello.
what we're using:
several Client PC's working on a terminal server which provides firefox and the maintenance service.
what happend:
firefox received an update. firefox was started.

Actual results:

after starting firefox a message box appeared which said that a new profile must be created. i could either create a new profile or just quit firefox.
firefox opened with a new profile und no bookmarks,history or whatever.
the original profile was still in the folder AppData\Roaming\Mozilla\Firefox\Profiles.

Expected results:

firefox should have kept the profile und user data after the update and shouldn't force a new proflle

Asked by Dave Townsend [:mossop]:' What was the install location of Firefox before and after the update? Looking at the compatibility.ini file from the old and new profiles would be helpful.'

Here is what I found in the compatibility.ini files from 3 different users.
FIRST USER:

[Compatibility]
LastVersion=72.0.2_20200117190643/20200117190643
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersAppDataRoamingMozillaFirefoxProfilesqw356cxa.default-release-5-1580717539211 // 05.02.2020

[Compatibility]
LastVersion=72.0.2_20200117190643/20200117190643
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersAppDataRoamingMozillaFirefoxProfiles6ni1j2sm.default-release-4 // 29.01.2020

[Compatibility]
LastVersion=70.0.1_20191030021342/20191030021342
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersAppDataRoamingMozillaFirefoxProfileszm7awtih.default-release-3 // vom 21.11.2019

SECOND USER:

[Compatibility]
LastVersion=72.0.2_20200117190643/20200117190643
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersuserxyAppDataRoamingMozillaFirefoxProfileshd2ynkbl.default-release-5 // vom 05.02.2020

[Compatibility]
LastVersion=72.0.2_20200117190643/20200117190643
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersuserxAppDataRoamingMozillaFirefoxProfilesz7h0h9ge.default-release-4 // 24.01.2020

[Compatibility]
LastVersion=70.0.1_20191030021342/20191030021342
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersuserxyAppDataRoamingMozillaFirefoxProfiles8fqlnptx.default-release-3 // 20.11.2019

[Compatibility]
LastVersion=70.0.1_20191030021342/20191030021342
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser
C:UsersuserxyAppDataRoamingMozillaFirefoxProfiles9mdt1oaw.default-release-2 // 11.11.2019

THIRD USER:

[Compatibility]
LastVersion=68.0.1_20190717172542/20190717172542
LastOSABI=WINNT_x86_64-msvc
LastPlatformDir=C:Program FilesMozilla Firefox
LastAppDir=C:Program FilesMozilla Firefoxrowser

mozilla maintainance service automaticly updates firefox on the terminal server for each user

These files do not seem correct. Can you please attach copies of them to the bug report rather than copying and pasting their contents?

Also for the future you can just re-open a bug when you're ready to provide the requested details.

Flags: needinfo?(manuel.rodrigues.gomes)
Attached file compatibility01.ini
Flags: needinfo?(manuel.rodrigues.gomes)
Attached file compatibility02.ini
Attached file compatibility03.ini
Attached file compatibility04.ini

several compability.ini files from affected users

Attachment #9126356 - Attachment mime type: application/octet-stream → text/plain
Attachment #9126357 - Attachment mime type: application/octet-stream → text/plain
Attachment #9126358 - Attachment mime type: application/octet-stream → text/plain
Attachment #9126359 - Attachment mime type: application/octet-stream → text/plain

What is the relationship between these compatibility.ini files?

Flags: needinfo?(manuel.rodrigues.gomes)

(In reply to Dave Townsend [:mossop] (he/him) from comment #7)

What is the relationship between these compatibility.ini files?

These files are from 3 different users who are affected by the problem.

Flags: needinfo?(manuel.rodrigues.gomes)

(In reply to manuel.rodrigues.gomes from comment #8)

(In reply to Dave Townsend [:mossop] (he/him) from comment #7)

What is the relationship between these compatibility.ini files?

These files are from 3 different users who are affected by the problem.

Are some from before the issue and some from after?

Flags: needinfo?(manuel.rodrigues.gomes)

(In reply to Dave Townsend [:mossop] (he/him) from comment #9)

(In reply to manuel.rodrigues.gomes from comment #8)

(In reply to Dave Townsend [:mossop] (he/him) from comment #7)

What is the relationship between these compatibility.ini files?

These files are from 3 different users who are affected by the problem.

Are some from before the issue and some from after?

Yes, these files are from yesterday.
The issue occurs (i think) after updates.
Some users have up to 6 different firefox profiles.
Everytime this happens the user loses his/her bookmarks, history...

Flags: needinfo?(manuel.rodrigues.gomes)

What I need to see is the compatibility.ini of the profile the user uses before the update and the compatibility.ini for the new profile that gets created after the update.

(In reply to Dave Townsend [:mossop] (he/him) from comment #11)

What I need to see is the compatibility.ini of the profile the user uses before the update and the compatibility.ini for the new profile that gets created after the update.

file 01 and 02 belong to the same user. file03 and 04 belong to another user.
file 02 and 04 are older than the others.

The priority flag is not set for this bug.
:mossop, could you have a look please?

For more information, please visit auto_nag documentation.

Flags: needinfo?(dtownsend)

Based on the compatibility files this isn't related to updates, suggests there is something broken with the profiles.ini files themselves.

Flags: needinfo?(dtownsend)
Priority: -- → P3

(In reply to Dave Townsend [:mossop] (he/him) from comment #14)

Based on the compatibility files this isn't related to updates, suggests there is something broken with the profiles.ini files themselves.

okay so what do you think is the best was to proceed now?

At the moment I have no way to reproduce this issue and there is no suggestion for what is triggering the problem. Ideally we'd be able to get a snapshot of a users state immediately before the issue occurs and then be able to reproduce that on a test machine but I'm not sure that that is very likely to work. There could be something specific to terminal services here that might be causing it but I can't see what it might be.

Yup, same behaviour, I would not assume the same cause, though the environment (TS) is similar.

This is really really bad for Firefox if we cannot find a way to fix this.

I guess I'm going to have to donate my time to perform the QA required. As it looks like the assigned staff does not have the resources required to do the job.

I can't really rollback and roll forward Firefox versions and play with user profiles on a production corporate network though. The users need to get thier work done.

Does Mozilla not have a terminal services test environment? How much would it cost you to set up a basic one? Perhaps we could have a fundraiser?

The Firefox build environment is such a pig though. Lib dependency hell, it takes days to configure and build all the projects it depends on.

I will stew on a way to fix this for you. We need Firefox now more than ever, and this issue will kill it for business use if we do not fix it.

Perhaps I could rent a cheap virtual server cluster and setup a build and test environment for you, then provide a login so you have the resources to work on this.

Any of these ideas seem helpful?

Actually, thinking on this some more, in order to properly test this, I would require control of your update servers or some way to replace them with identical functionality in a custom FF build.

Is that possible? Is the code and config for your update servers available?

Unless you have a staging environment for your systems you probably just need to wait for the next Firefox update. Ideally I would like to see the entire contents of %APPDATA%\Mozilla\Firefox before an update and then after running the updated version and seeing the profile lost. Should be straightforward to do with a test user account.

You are still asking me to do your software QA on my live production corporate network. I'm sorry but that's not acceptable to me (nor should it be to you either).

Unless Mozilla can get it together (and based on what I see lately, it is not going to happen, they're concerned with marketing data, not the code), and if I want to continue to use firefox, it is obvious that I will have to fix it myself.

Yes, I will have to build an expensive and time consuming staging environment (build, update servers, custom ff build, clients) in order to fix it for you.

But you see, I have the rest of my business to run. When will I sleep? My clients will not pay me for this.

(In reply to ReidTech from comment #21)

You are still asking me to do your software QA on my live production corporate network. I'm sorry but that's not acceptable to me (nor should it be to you either).

I'm afraid that for issues that only appear in non-standard environments we often have to rely on the reporters helping us gather enough information to be able to understand what is happening.

Yes, I will have to build an expensive and time consuming staging environment (build, update servers, custom ff build, clients) in order to fix it for you.

There is no need to test this with custom builds of Firefox or custom update servers. In fact I'd prefer it if you didn't, that only serves to add noise to the data.

Fair enough. So you'd rather I wait.

Here's why I have a problem with that: search the bug tracker for profile bugs, there are dozens of them, years old, that are not even assigned, yet there are "bugs" in this tracker that are about user data collection (nothing to do with FF core code) that have been assigned.

If Mozilla is concerned about market share and competition, all the data collection in the world will not help if the product quality is bad.

Build a product of superior quality and it will dominate.

So why is this relevant in this discussion? You mention "non-standard environments".

Well, MS Terminal Services is now a "standard environment". Even before COVID companies have been moving to hosted remote desktops like crazy, now that COVID requires everybody to work from home for an indefinite period of time (permanently?) - well...

If you do not consider MS Terminal Services a standard environment, you are going to lose corporate market share.

The MUA (or whatever) user data collection will not help. Only fixing the code will.

And so I do not have the faith at this point that Mozilla can see the product anymore. The beancounters have too much control. Code needs to be controlled by coders, not managers concerned with social issues and sales data.

Look at it...

  • Pocket (nobody wants it)
  • TRR (hijacking corporate DNS to report it to cloudflare - completely unacceptable)
  • Firefox accounts (nobody wants them)

Just build solid code and forget all the silly cruft. Focus on the actual browser functionality.

I know its silly to make this speech to you. The Mozilla leadership really needs to hear it before they finish killing Firefox. Maybe they will see it?

I will stew on whether to keep my clients on Firefox or to move to Edgium (Chrome is completely unacceptable for science or business). I'd really rather not - I want Firefox to be a solid and free as in speech product. I'm not sure if I will do it myself or not. I have a few major projects to finish.

Can you clarify exactly what your server is running and how applications are exposed to users? The windows folks are telling me that the product "Terminal Services" was renamed a number of years ago and Remote Desktop Services encompasses a number of different things.

Flags: needinfo?(brian)

OS: Windows Server 2012 Standard
Application: Remote Desktop Services

Remote Desktop Services primary function is to run an rdp (remote desktop protocol) server (Remote Desktop Session Host). The Windows Remote Desktop is similar to VNC or Xterm but different. What it does is provide a remote UI for a local user session. The Applications are delivered like any other, it runs on the computer/server in a user session. The windows server runs a service invoked thus:

C:\Windows\System32\svchost.exe -k termsvcs

The service listens on port 3389 for rdp client connections. When a client connects with the rdp client (mstsc.exe), they are presented a windows login, and once logged in are presented with their full user desktop in a window.

Imagine you are connecting to your (assuming) Ubuntu Gnome desktop that is running VNC. When you login, you see the desktop and UI. The user experience is essentially the same.

Remote Desktop Services may do other undocumented things, but what it does is run a Remote Desktop Session Host. Any other functions it has are ancillary to the actual remote desktop service.

Flags: needinfo?(brian)
OS: Unspecified → Windows
Summary: creates new profile after update → creates new profile after update. terminal services related?
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: