Autostarts but doesn't run, will run on restart

ember1205

I just edited the run-start.sh script and inserted a ten second sleep before the pgrep command. It’s a bit “extreme”, but should hopefully demonstrate if that’s the issue.

WIthout the sleep, not detecting X simply moves on to a different test instead of trying again at least once before moving on.

sdetweil

@ember1205 yes, adding a sleep would ‘fix’ it

if only we knew how LONG to sleep… working on it…

if X is not running then we just launch serveronly and don’t look at its output

sdetweil

@sdetweil try 1 second… sleep 1

ember1205

@sdetweil

Sleep 10 allows the browser to correctly launch. I will drop to 1 second and validate that it will work or not (if not, I will tweak and tune).

Even with the sleep function ensuring that it allows X to finish initializing, chromium does not start correctly unless I wipe the ~/.config/chromium directory when I shut down.

sdetweil

@ember1205 people are constantly fiddling with stuff, you never know what you’re gonna get anymore.

ember1205

@sdetweil

Would there be more value in coding in a check for to see if it’s Raspbian or possibly even bullseye and adding a two second sleep if there is?

raspberry_pi=$(grep ID /etc/os-release | cut -f2 -d=)
if [ "$raspberry_pi." == "raspbian" ] ; then
   sleep 2
fi
xorg=$(pgrep Xorg)

Maybe there’s a better string to search for, but it might be a stop-gap that would help?

sdetweil

@ember1205 i don’t think thats right… as we could have the same problem on any other system…

ember1205

@sdetweil

Correct syntax would need to be:

  raspberry_pi=$(grep ^ID= /etc/os-release | cut -f2 -d=)
  if [ "$raspberry_pi." == "raspbian" ] ; then
   sleep 2
  fi

The RPi 3B+ is a little slower to initialize on Bullseye versus Buster. Any time a device is running a slightly heavier OS and needs a little more time to finish booting up, I could see this occurring. The question is, do you solve it with a fix that impacts everyone (“sleep 2” or similar), test for particular OS/HW combo’s, or is there another process to look for that would better indicate when the system should be finished initializing?

Right now, it looks for X and just “moves on” when it isn’t found and assumes some sort of mistake or something in the config. Great as a fail-safe, but creates a new occasional issue to be dealt with.

sdetweil

@ember1205 that test is for when MM is running under docker (with no X running) but using the docker host’s xorg for viewing…

ember1205

@sdetweil

The test I’m referring to checks to see if X is running. If not, it moves on to test for Lubuntu. If that also fails, it assumes server only and moves on.

I do recall seeing a specific Docker test at the very beginning of the script, but that’s a fair bit earlier than where I’ve added this check (which is right before the “pgrep Xorg” line).