Jump to content
Enpass Discussion Forum

Enpass 6 beta for Linux is now available


Gajender Singh
 Share

Recommended Posts

Same issue here (Chrome and Firefox). I was also wondering if there is also an extension update ?

Settings->Browser->Review Browsers is empty.

As a side note, I am a developer and Linux server administrator, ask me anything technical to help you find the issue.

Ubuntu 18.10 (Gnome Shell)
Enpass 6.0.0 (214)
Firefox Extension 6.0.0.56b
Chrome Extension  6.0.0.56

Edited by Lox_
Link to comment
Share on other sites

Hi @Lox_,

We tried to reproduce this issue with a fresh virtual machine for Ubuntu 18.10 with default configurations. All browser extension are working fine. There must be something we are missing here.

How an extension connects with Enpass is not radically different from version 5. Enpass runs a websocket server on 127.0.0.1 on one of the port (10391 - 10395). An extension tries to connect to one of these port, if it couldn't find the server or Enpass couldn't determine connecting browser pid it fails. Please share following details:

1. Output of `lsof -i:10391`
2. Are you using any proxy? If yes, please set it to ignore 127.0.0.1. Also, adjust any firewall rule accordingly.
3. Are multiple instance of Enpass running in different userspace (a multiuser system)?
4. Chrome and Firefox version and source of installation (ubuntu repo, snap or flatpack)

Your time and effort are greatly appreciated.

Link to comment
Share on other sites

10 hours ago, Lox_ said:

Same issue here (Chrome and Firefox). I was also wondering if there is also an extension update ?

Settings->Browser->Review Browsers is empty.

As a side note, I am a developer and Linux server administrator, ask me anything technical to help you find the issue.

Ubuntu 18.10 (Gnome Shell)
Enpass 6.0.0 (214)
Firefox Extension 6.0.0.56b
Chrome Extension  6.0.0.56

@Vinod Kumar I actually have two machines, both have authenticated browser, but on one machine it works, on another it still shows error and extension does not work.

$ lsof -i:10391
COMMAND  PID     USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5261 <username>   30u  IPv4  41930      0t0  TCP *:10391 (LISTEN)

I'm using Firefox 63.0.3 installed from official Fedora repos (@updates)

Edited by Alex Gurenko
Link to comment
Share on other sites

14 hours ago, Vinod Kumar said:

Hi @Lox_,

We tried to reproduce this issue with a fresh virtual machine for Ubuntu 18.10 with default configurations. All browser extension are working fine. There must be something we are missing here.

How an extension connects with Enpass is not radically different from version 5. Enpass runs a websocket server on 127.0.0.1 on one of the port (10391 - 10395). An extension tries to connect to one of these port, if it couldn't find the server or Enpass couldn't determine connecting browser pid it fails. Please share following details:

1. Output of `lsof -i:10391`
2. Are you using any proxy? If yes, please set it to ignore 127.0.0.1. Also, adjust any firewall rule accordingly.
3. Are multiple instance of Enpass running in different userspace (a multiuser system)?
4. Chrome and Firefox version and source of installation (ubuntu repo, snap or flatpack)

Your time and effort are greatly appreciated.

» lsof -i:10391
COMMAND    PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass    8707  lox   48u  IPv4  70877      0t0  TCP *:10391 (LISTEN)
Enpass    8707  lox   59u  IPv4  84833      0t0  TCP localhost,:10391->localhost,:36776 (ESTABLISHED)
firefox   9438  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)
chrome-gn 9741  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)
gjs       9985  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)

No proxy

Only a single instance is running

» ps aux | grep Enpass
lox       8707  0.3  1.0 1524564 159364 tty2   Sl+  08:44   0:02 /opt/enpass/Enpass -minimize

I have automatic updates so firefox is version 63.0.3 and chrome 70.0.3538.110

Link to comment
Share on other sites

Ugh. So I finally managed to get it working as well.

Can't say 100% which of the following finally did it but here's what I did:
  * Remove extension from within chromium (note the id for later)
  * Close chromium
  * If still exist, remove both `.config/chromium/Default/Local Extension Settings/<id>` and `~/.config/chromium/Default/Extensions/<id>`
  * Remove everything extension related (use find and/or grep) from within `~/.cache/chromium/` (I nuked the whole chromium folder)
  * Start chromium
  * Install extension

I think removing the cache folder was what finally solved it.

Link to comment
Share on other sites

Hi @Lox_

There is something odd in lsof output. All three process firefox, chrome-gn and gjs are connceted with Enpass with same peer port (36776) and FD. Peer ports should be different like

localhost:~$ lsof -i:10391
COMMAND  PID  USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
firefox 3140 vinod  110u  IPv4  56432      0t0  TCP localhost:48872->localhost:10391 (ESTABLISHED)
chrome  3572 vinod  260u  IPv4  54522      0t0  TCP localhost:48858->localhost:10391 (ESTABLISHED)
Enpass  4289 vinod   40u  IPv4  53849      0t0  TCP *:10391 (LISTEN)
Enpass  4289 vinod   43u  IPv4  55788      0t0  TCP localhost:10391->localhost:48858 (ESTABLISHED)
Enpass  4289 vinod   55u  IPv4  55825      0t0  TCP localhost:10391->localhost:48872 (ESTABLISHED)

What is this process gjs and why it is connecting to Enpass?

 

Link to comment
Share on other sites

Hi @Tadly,

There seems some problem with extension sandbox cache which is messing up authorization with Enpass app. @Alex Gurenko 's workaround also confirms that. We are looking into it.

On a different note, is Enpass assistant (mini enpass window) working on your i3wm setup now? We adjusted some window flags per Desktop/Window manager to avoid focus problems.

Link to comment
Share on other sites

@Vinod Kumar Had to find my old post to remember... ':D

Seems to be working fine now, yes.
If you'd like to do me (and possibly others) a huge favor, adjust the instance-name for WM_CLASS to be different for the assistant:)
Currently it's (for both the main app and the assistant):

WM_CLASS(STRING) = "Enpass", "Enpass"

Something like (for the assistant):

WM_CLASS(STRING) = "Enpass-Assistant", "Enpass"

would be amazing (for automation purposes that is)

 

Remaining issues:
1. Within the assistant, Arrow down has to be hit TWICE for the first entry to be selected.
Where is the focus after the first press? I don't know hence it can't be important :)

2. Within the assistant, after pressing arrow right on a highlighted item, it opens (so far so good).
Within the item I am still unable to navigate up/down using my arrow keys.
This would be necessary to copy values to the clipboard (For me TOTP is often the case)

3. Assistant closes instantly after loosing focus.
If I accidentally moved the cursor (touchpad) of off the window the window would just go away and I have to start over.
This behavior is likely specific to window managers as it depends on how focus tracking behaves.
On my setup a mouse-over focuses a window.
As the assistant is OVER my browser (floating) this can trigger rather 
quickly due to the top-left corner being anchored to the cursor position.
=> Move one pixel up/left and the window is gone.
Either add a close delay (max 1 sec.) or open the assistant centered on the cursor (Probably a nicer implementation)

4. SCROLLING!
I already mentioned it in my other post. Don't do custom scrolling. PLEASE!
I doubt (correct me if I'm wrong) Qt forces you to do so = Don't!
It feels hecking awful on linux (touchpad) and it feels bad on windows (scroll-wheel).

 

These are the most pressing things from where I stand (and what I found from the little playing I did).
As someone with very high standards there are lots more things I could ask to improve but I don't want to push my luck here :D
If you guys are interested in a bigger list of things to improve/add I could try to find some time and do another write up but for now
that's all I got.

Unlike with my previous write-up I did actually migrate to this version and started using it.
So I don't deem it unusable anymore which is good :)

Edited by Tadly
Link to comment
Share on other sites

10 hours ago, Vinod Kumar said:

There is something odd in lsof output. All three process firefox, chrome-gn and gjs are connceted with Enpass with same peer port (36776) and FD. Peer ports should be different like


» lsof -i:10391
COMMAND    PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass    8707  lox   48u  IPv4  70877      0t0  TCP *:10391 (LISTEN)
Enpass    8707  lox   59u  IPv4  84833      0t0  TCP localhost,:10391->localhost,:36776 (ESTABLISHED)
firefox   9438  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)
chrome-gn 9741  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)
gjs       9985  lox  143u  IPv4  85268      0t0  TCP localhost,:36776->localhost,:10391 (ESTABLISHED)

What is this process gjs and why it is connecting to Enpass?

 

Hello @Vinod Kumar

Both were there long before and working well alongside Enpass 5. Anyway I removed both apps and rebooted.
I also uninstalled both, firefox and chrome, extensions and closed the browsers, reopened them and reinstalled the extensions, and finally rebooted.

After a clean reboot I have:

COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5674  lox   48u  IPv4  59720      0t0  TCP *:10391 (LISTEN)

When I open Firefox I have:

COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5674  lox   48u  IPv4  59720      0t0  TCP *:10391 (LISTEN)
Enpass  5674  lox   60u  IPv4  80261      0t0  TCP localhost,:10391->localhost,:47982 (ESTABLISHED)
firefox 7496  lox  196u  IPv4  81263      0t0  TCP localhost,:47982->localhost,:10391 (ESTABLISHED)

As soon as I click the extension icon and just before the new tab opens up it changes to:

COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5674  lox   48u  IPv4  59720      0t0  TCP *:10391 (LISTEN)

I closed Firefox, opened Chrome and got:

COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5674  lox   48u  IPv4  59720      0t0  TCP *:10391 (LISTEN)
Enpass  5674  lox   60u  IPv4  90463      0t0  TCP localhost,:10391->localhost,:48248 (ESTABLISHED)
Enpass  5674  lox   69u  IPv4  89923      0t0  TCP localhost,:10391->localhost,:48250 (ESTABLISHED)
chrome  8289  lox  274u  IPv4  90462      0t0  TCP localhost,:48248->localhost,:10391 (ESTABLISHED)
chrome  8289  lox  279u  IPv4  89922      0t0  TCP localhost,:48250->localhost,:10391 (ESTABLISHED)

As soon as I click the extension icon and just before the new tab opens up it changes to:

COMMAND  PID USER   FD   TYPE DEVICE SIZE/OFF NODE NAME
Enpass  5674  lox   48u  IPv4  59720      0t0  TCP *:10391 (LISTEN)
Enpass  5674  lox   69u  IPv4  89923      0t0  TCP localhost,:10391->localhost,:48250 (ESTABLISHED)
chrome  8289  lox  279u  IPv4  89922      0t0  TCP localhost,:48250->localhost,:10391 (ESTABLISHED)

It still fails in both cases. Can't this just be related to upgrading from Enpass 5? This is basically all I have done: upgraded everything to 6. I can't remember changing anything since then, except automatic updates. Isn't there any log file to watch?

Regards

Edited by Lox_
Link to comment
Share on other sites

  • 2 weeks later...

Hi @Tadly,

Please try out new beta. We have incorporated some of the changes you asked.

On 12/12/2018 at 7:49 PM, Tadly said:

If you'd like to do me (and possibly others) a huge favor, adjust the instance-name for WM_CLASS to be different for the assistant:)

We can't. Both main and assistant windows are part of same executable instance and according to specs, windows of same app should have same WM_CLASS. However, you can use alternative x atoms for automation purposes. Like

WM_NAME(STRING) = "Enpass Assistant"
_NET_WM_NAME(UTF8_STRING) = "Enpass Assistant"

 

On 12/12/2018 at 7:49 PM, Tadly said:

1. Within the assistant, Arrow down has to be hit TWICE for the first entry to be selected.
Where is the focus after the first press? I don't know hence it can't be important :)

First down press shifts focus from search to list and second press is handled by list.

On 12/12/2018 at 7:49 PM, Tadly said:

2. Within the assistant, after pressing arrow right on a highlighted item, it opens (so far so good).
Within the item I am still unable to navigate up/down using my arrow keys.
This would be necessary to copy values to the clipboard (For me TOTP is often the case)


Yes, this is a unfinished task. Coming in subsequent updates. Till than you can same keyboards shortcuts as in main app for detail page.

On 12/12/2018 at 7:49 PM, Tadly said:

3. Assistant closes instantly after loosing focus.
If I accidentally moved the cursor (touchpad) of off the window the window would just go away and I have to start over.
This behavior is likely specific to window managers as it depends on how focus tracking behaves.
On my setup a mouse-over focuses a window.
As the assistant is OVER my browser (floating) this can trigger rather 
quickly due to the top-left corner being anchored to the cursor position.
=> Move one pixel up/left and the window is gone.
Either add a close delay (max 1 sec.) or open the assistant centered on the cursor (Probably a nicer implementation) 

I have sneaked a environment variable till we have a better solution. Just set the delay in seconds

ENPASS_ASSISTANT_AUTOHIDE_DELAY=1

 

On 12/12/2018 at 7:49 PM, Tadly said:

4. SCROLLING!
I already mentioned it in my other post. Don't do custom scrolling. PLEASE!
I doubt (correct me if I'm wrong) Qt forces you to do so = Don't!
It feels hecking awful on linux (touchpad) and it feels bad on windows (scroll-wheel). 

This is a qt bug with High precision touchpads. We were not using any custom scrolling, but now we are. Scrolling should feel better now.

Cheers:)

Link to comment
Share on other sites

1 hour ago, Vinod Kumar said:

We can't. Both main and assistant windows are part of same executable instance and according to specs, windows of same app should have same WM_CLASS. However, you can use alternative x atoms for automation purposes.

The WM_NAME atom was the same for both previously right? I could have sworn I checked that. Either way, that's enough for me, thanks!
P.s. What i was revering to by saying WM_CLASS is what `xprop` shows. The output of `xprop` includes the "window instance" (fist element) and "window class" (second element).
The window class should be the same, yes. The instance (as far as I'm aware) can be a different one.

1 hour ago, Vinod Kumar said:

First down press shifts focus from search to list and second press is handled by list

I know this sounds lazy, but can you have it jump straight to the first entry?
I'm a dev myself and understand your argument (from a technical POV it "makes sense") but the user doesn't care especially if it's one who likes to be efficient :)

1 hour ago, Vinod Kumar said:

Yes, this is a unfinished task. Coming in subsequent updates. Till than you can same keyboards shortcuts as in main app for detail page

Glad to hear it! Hope this will be implemented soon :)

1 hour ago, Vinod Kumar said:

I have sneaked a environment variable till we have a better solution. Just set the delay in seconds

Thanks but there's a slight oversight. The delay works but if I refocus the window (so I move the cursor away and within one second back to it) it still closes after one sec.
If it should gain focus again, it shouldn't close :)

1 hour ago, Vinod Kumar said:

This is a qt bug with High precision touchpads. We were not using any custom scrolling, but now we are. Scrolling should feel better now.

Good lord! I'm sorry than :/
How come this hasn't been fixed yet? If you have a bug report link for this specific bug I'd love for you to link it :)
Anyways... It's working much better now! Thanks a bunch!

Cheers and merry Christmas!

Link to comment
Share on other sites

Can somebody help me? After installation it’s not possible to launch the Enpass 6 app. I already tried to remove it and install again but every time having the same issue. If I remove Enpass 6 and install Enpass 5 again, there are no such problems. 

I’m running it on a Linux Mint 18.3 machine. 

Link to comment
Share on other sites

Im using the latest Enpass beta from the AUR (https://aur.archlinux.org/packages/enpass-beta-bin/)
Also I'm using the latest Android Beta, with Google Drive Sync turned on.
If I want to connect Enpass on my Linux machine with my Google Drive Account, I get the following error:image.thumb.png.258080d7eda74dce4c02414a4ed7d86d.png

Using Enpass Beta on a second Android device with Google Drive is working fine.
Is there anything I can do?

Edited by Aragur
Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
 Share

×
×
  • Create New...