Terminator crashes 1-10 seconds after opening

[RoelVdP]

$ cat /etc/*release* | grep -E 'DESC|VER.*ID'
DISTRIB_DESCRIPTION="Ubuntu 20.04.1 LTS"
VERSION_ID="20.04"

Already tried:

1. sudo apt -y purge terminator && sudo apt -y install terminator -> Same result
2. sudo apt -y purge terminator && git clone https://github.com/gnome-terminator/terminator.git && {install deps} && {compile & install}
3. Tried running with another older Python version (normally it uses 3.8), but that produced symbol missing.

Same result after all attempts to fix

Issue started happening today with no real clear change in the setup other then perhaps creating some groups and then closing/re-opening terminator.

Grabbing the window by the title bar (mouse left-click + drag) seems to make the issue produce immediately.

Also, when it crashes, the process keeps running though the terminal window itself is gone. No output in dmesg and not sure where else to look for logs. Killing the process (-9) and re-opening has no effect, issues reproduces again and again after a few seconds.

Please assist, I use terminator very often for work.

[RoelVdP]

It seems to be related to a particular configuration. i.e. have a particular config in place (4 windows, auto-start a script), then open config, don't save anything (though config file seems to get date/time update), and from that moment the crashes happen. Removing the config also stops the crashes.

[RoelVdP]

Completely re-configured the environment, and it continues to work without issues. In conclusion, it must be something in the config file which is updated when the configuration window opens and closes, or when terminal groups are updated, and then perhaps in combination with some backword compatibility or similar. I can perhaps upload the crashing config file.

[RoelVdP]

Steps to reproduce (steps 8 and onwards are cleanup)

1. close all terminator windows, check with ps -ef | grep terminator there are no live processes
2. cd ~/.config && mv terminator terminator.OLD
3. mkdir terminator && cp ~/Downloads/config ./terminator/config
   The config file is @ https://gist.github.com/RoelVdP/26d59a27bbf2809819b608902dd23f43
4. open terminator, click on titlebar, keep in left mouse, drag down
5. terminator should crash < # The [bug]
6. check with ps -ef | grep terminator and note process is still there
7. kill -9 pid where pid is the pid from the command in Implement support for Python 3 #6
8. cd ~/.config && mv terminator terminator.BROKE && mv terminator.OLD terminator
9. restart your own terminator
   Also, the ~/s script mentioned in the config file is simply a ssh to google cloud I use for work.

[mattrose]

I'll take a look at this when I get the chance, but if you could run terminator at another command prompt by opening up gnome-terminal and running terminator -d and attaching the output here that would make things go quicker.

Also, can you run terminator -v to tell us what version you're on?

[derenv]

also had this problem (with terminator 2.0.1 on Ubuntu 20.10), renaming my config file and creating a new one (setting new preferences) seemed to solve. These are my config files: https://gist.github.com/derenv/3a9ece6dc7980c4e0afe773bd3b4ca7f

I recently upgraded the terminator version, could that be affecting it? The crash tended to crash the entirety of GNOME shell too, especially if loads of other programs are running.

[mattrose]

Tried it with both config files and they both worked fine on my 20.04.1 VM. @RoelVdP's config had a ~/s command that may be causing the problem, as it is set to restart when it fails, so on my VM it just constantly restarts without doing anything. If this command can cause CPU usage on start it might be causing the issue. Anyways, attache terminator -d output and that should tell us if anything is going wrong within terminator

[derenv]

https://gist.github.com/derenv/5df2c1b13455e0cd1b88c7d194f06e20

[mattrose]

Hmm, I don't see anything unusual in that debug output. Did you close it yourself, or did it close through some other method?

[derenv]

once it crashes terminator restarts (with the 1-10 second delay) and a new window spawns then works normally, at which point i closed it myself.

[mattrose]

Hmm, I don't see it crashing at all. When the new window spawns, is there new debug output in the window that you originally launched it from? I wish I could reproduce.

[derenv]

nah, it seems to maintain the debug output of the initial window (the quit proccess part at the end of that gist is me closing the new window).

this has been bugging me (ha) for a while im just glad a new config file fixes it.

[mattrose]

The only thing I can think of doing from here is going through the old config file line-by-line and seeing which line causes the problem.

[derenv]

Found the culprit!

maximised = True

@RoelVdP can you try to remove that line from yours and see if that fixes it for you too?

[mattrose]

Thank you so much for this!

[RoelVdP]

No, still crashes when I remove the 3x 'maximised = True'. It also has nothing to do with the '~/s', that is just a script and it does not loop. I removed 'use_custom_command' as well as 'custom_command` from the config file completely, to be sure and it still crashes all the same. 1-3 seconds before crash.

[RoelVdP]

$ terminator -v
terminator 2.0.1

And end of log:

Config::set_profile: Config::set_profile: Changing profile to default
Config::connect_gsetting_callbacks: GSetting connects for system changes
noclass::parse_options: OptionParse::parse_options: command line options: {'version': None, 'maximise': None, 'fullscreen': None, 'borderless': None, 'hidden': None, 'forcedtitle': None, 'geometry': None, 'command': None, 'config': '/home/me/.config/terminator/config', 'configjson': None, 'execute': None, 'working_directory': None, 'forcedicon': None, 'role': None, 'layout': 'default', 'select': None, 'profile': None, 'nodbus': None, 'debug': 1, 'debug_classes': None, 'debug_methods': None, 'new_tab': None, 'dummy': None}
Terminator::__init__: Borg::__init__: Preparing borg state for Terminator
ConfigBase::load: ConfigBase::load: config already loaded
Config::set_profile: Config::set_profile: Changing profile to default
Config::connect_gsetting_callbacks: GSetting connects for system changes
ConfigBase::get_item: ConfigBase::get_item: broadcast_default found in globals: group
ConfigBase::load: ConfigBase::load: config already loaded
Config::set_profile: Config::set_profile: Changing profile to default
Config::connect_gsetting_callbacks: GSetting connects for system changes
ConfigBase::get_item: ConfigBase::get_item: dbus found in globals: True
function::with_proxy: dbus client call: new_window_cmdline
function::with_proxy: dbus client call: new_tab_cmdline
function::with_proxy: dbus client call: new_window
function::with_proxy: dbus client call: new_tab
function::with_proxy: dbus client call: hsplit
function::with_proxy: dbus client call: vsplit
function::with_proxy: dbus client call: get_terminals
function::with_proxy: dbus client call: get_window
function::with_proxy: dbus client call: get_window_title
function::with_proxy: dbus client call: get_tab
function::with_proxy: dbus client call: get_tab_title
DBusService::__init__: Borg::__init__: Preparing borg state for DBusService
DBusService::prepare_attributes: Checking for bus name availability: net.tenshu.Terminator21a9d5db22c73a993ff0b42f64b396873
DBusService::prepare_attributes: bus name unavailable: net.tenshu.Terminator21a9d5db22c73a993ff0b42f64b396873
noclass::<module>: Unable to become master process, operating via DBus
noclass::<module>: Requesting a new window

[mattrose]

can you make sure you're passing -u, so terminator -d -u just so that you're getting debug output from the command.

[mattrose]

Closing