1195137 - T-Bird 38.1.0 does not start because a previous session was not shut down.

Status: RESOLVED INCOMPLETE

(Thunderbird :: Untriaged, defect)

Description

[bruce korb]

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/44.0.2403.155 Safari/537.36

Steps to reproduce:

I got thunderbird into a wedged state somehow or another (different problem), so the windows all disappeared leading me to think it was not running.  I could not restart it because it was already running.


Actual results:

The new instance exited without leaving a clue about the problem.  I had to spend a half hour doing research to get me to where I could "ps -C thunderbird" to see that it was running and then shoot it down.


Expected results:

A window should pop up asking whether to simply exit or to kill the running tbird.  i.e. the window should offer to "kill -TERM $tbird_pid" and then start up tbird.

Comment 1

[Wayne Mery (:wsmwk)]

Please try 38.2.0 https://www.mozilla.org/en-US/thunderbird/

Comment 2

[bruce korb]

I looked in the bug reports and didn't see a similar one.
If you believe you've fixed it, then I'll believe in you.
I will wait for it, but not by deliberately wedging t-bird
and seeing if I get a message upon restart.  "It's too hard."
Please close this if you've already fixed this issue.

Comment 3

[Wayne Mery (:wsmwk)]

If I understand correctly, A. your previous session did not shut down, and your next session appeared to have not started.  These aspects are indeed reported in other bugs.  But A. should not be happening in version 38 because there is a shut hang detector in version 38.

In any event, we can't test your environment. If the problem contineus the ball is in your court to try safe mode/provide more details about your environment

Comment 4

[bruce korb]

Thank you, Wayne.  "These aspects are indeed reported in other bugs" I did not select the right search term that brought them forward.  Bugzilla searching leaves a lot to be desired.

Since the hang is a rare occurrence (first time, actually), I didn't bother with a bug report for that.  I shot it down and all was well.  Still, when trying to start a session when there is an existing, windowless, non-communicative session still going, there needs to be a window that explains the situation.  Silently exiting is a really poor response.  By closing this, you are saying that 38.2 now detects the comatose session and alerts the user, correct?

Comment 5

[bruce korb]

"But A. should not be happening in version 38 because there is a shut hang detector in version 38."

I guess another bug:  the "shut hang detector" failed.