Bug 31192 - Thunderbird very slow then crashes even with --safe mode after last urpmi --auto-update
Summary: Thunderbird very slow then crashes even with --safe mode after last urpmi --a...
Status: NEW
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: 8
Hardware: All Linux
Priority: Normal major
Target Milestone: ---
Assignee: Nicolas Salguero
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-11-26 09:07 CET by Maat
Modified: 2022-11-28 10:15 CET (History)
2 users (show)

See Also:
Source RPM: thunderbird-102.5.0-1.mga8.src.rpm
CVE:
Status comment:


Attachments
Htop output (58.22 KB, image/png)
2022-11-26 17:13 CET, Maat
Details
lenovo laptop w/8GB ram running TB and FF Mga8 for comparison (167.63 KB, image/png)
2022-11-26 17:57 CET, Barry Jackson
Details

Description Maat 2022-11-26 09:07:45 CET
Description of problem:

Since last urpmi --auto-update thunderbird is so slow that it's unusable.
I tried to run it with --safe-mode to disable extensions and themes : same result
I dit let it run all the night just in case it needed to "process" somehow it's database(s). At the end of the night it coredumped wit this message :

$ thunderbird --safe-mode --jsconsole
ATTENTION: default value of option mesa_glthread overridden by environment.
out of memory: 0x0000000000000048 bytes requested
Exiting due to channel error.
Crash Annotation GraphicsCriticalError: |[C0][GFX1-]: Receive IPC close with reason=AbnormalShutdown (t=2130.34) Erreur de segmentation (core dumped)

At least one other case of this problem is reported on Mageia Online french forum : https://www.mageialinux-online.org/forum/topic-29468+thunderbird-crash.php


Version-Release number of selected component (if applicable): thunderbird-102.5.0-1.mga8


How reproducible:

Always : I just run thunderbird on my laptop and boom

Steps to Reproduce:
1. Open a console
2. Run thunderbird
3.
Comment 1 Maat 2022-11-26 09:10:12 CET
I also tried to reboot the computer, change kernel,  check the js console in search of noticeable errors without success.

Seems it's a "low level" problem...
Comment 2 Dave Hodgins 2022-11-26 14:49:49 CET
It may be only impacting some configurations or usage. I started thunderbird,
and using htop to monitor it's memory usage. It goes up and down even though it's currently in the background, but so far keeps returning to the same amount
of ram used.

With 16GB of ram on my system it would take a while before any memory leak
would force the system to start swapping. With 32GB of swap on my ssd drive,
quite a while before it would crash.

Assigning to Nicolas, though I doubt we'll be able to do anything other
then wait for the next thunderbird esr update.

Maat, how much ram/swap is in your system, and what architecture (from
"rpm -q -i thunderbird|grep Arch").
 
Use "free -m" to get ram/swap available/used.

CC: (none) => davidwhodgins
Assignee: bugsquad => nicolas.salguero

Comment 3 Maat 2022-11-26 17:05:39 CET
Hi Dave,

So for the 1st question :

$ LAN=C rpm -q -i thunderbird|grep Arch
Architecture: x86_64

for the second :

$ LANG=C free -m
               total        used        free      shared  buff/cache   available
Mem:           31809        4614       23926        1091        3268       25659
Swap:          59903           0       59903

Bonus :

ps -eFl | grep thunderbird
4 R maat       83319   77111 99  80   0 - 1541161 -    3154984 4 16:59 pts/1    00:04:23 thunderbird --safe-mode --jsconsole
0 S maat       86237   83319  0  80   0 - 606509 do_sys 97200  7 16:59 pts/1    00:00:00 /usr/lib64/thunderbird/thunderbird -contentproc -childID 1 -isForBrowser -prefsLen 165607 -prefMapSize 380819 -safeMode -parentBuildID 20221116090026 -appDir /usr/lib64/thunderbird 83319 tab
Comment 4 Maat 2022-11-26 17:13:07 CET
Created attachment 13534 [details]
Htop output

Bonus 2 : the output of htop.

One TB process is eating between 100% and 130% on one Core line.

but the laptop is far from its RAM limit.
Comment 5 Dave Hodgins 2022-11-26 17:42:28 CET
Looks like in addition to a memory leak, it's using excessive cpu.
See if https://rainbow.chard.org/2013/02/19/thunderbird-high-cpu/ helps.
Comment 6 Barry Jackson 2022-11-26 17:57:23 CET
Created attachment 13535 [details]
lenovo laptop w/8GB ram running TB and FF Mga8 for comparison

Added htop from my laptop for comparison as I was reading this and decided to check.
There are several pages of thunderbird processes but not doing much other than occasional short bursts of activity on one core.
I have only one 'remove duplicates' addon IIRC.

CC: (none) => zen25000

Comment 7 Maat 2022-11-26 22:32:05 CET
(In reply to Dave Hodgins from comment #5)
> Looks like in addition to a memory leak, it's using excessive cpu.
> See if https://rainbow.chard.org/2013/02/19/thunderbird-high-cpu/ helps.

Il have already the higher value for mail.db.idle_limit
Comment 8 Maat 2022-11-26 23:53:51 CET
I tested 102.5.0 and 105.4.2 from https://www.thunderbird.net/ : the behavior is slightly more responsive but the process eating nearly one CPU line is there

Quite different : version  91.13.1 from https://www.thunderbird.net/ starts fast and seems to work as expected. But this one eats a lot more RAM. 

(Nota: my profile is really big: 100+G with 10+ IMAP accounts and lots of archives)

So 1st conclusion : the problem is still in the upstream built version = It's not a packaging issue.

2nd conclusion : it might be global to all 102.x but i did not check versions earlier than 102.4.2
Comment 9 Dave Hodgins 2022-11-27 05:45:43 CET
Probably best to report it at
https://www-archive.mozilla.org/support/thunderbird/bugs
using the latest build.

Working directly with them will hopefully make solving it faster.
Comment 10 Maat 2022-11-27 08:56:09 CET
Insightful, thanks Dave.
Comment 11 Maat 2022-11-28 10:15:01 CET
After further tests 91.13.1 version ends also in the same problem... it just needed a night to show up.

Currently i'm testing with  91.13.1 BUT i booted on 5.18.11-desktop-1.mga8

And it seems better.

Could it be an issue with something in the new 6.x kernel ?

Note You need to log in before you can comment on or make changes to this bug.