Skip to content

debian/bluebinder.service: restart less frequently, but for more times

Ratchanan Srirattanametrequested to merge
personal/peat-psuwit/increase-restart into ubports/latest

Sometimes, the HIDL service isn't really ready yet despite the service's process already running (as detected by bluebinder_wait.sh), which leads to bluebinder failling. When this occur, by default systemd will restart the service after 100 ms up to 5 times, which doesn't help in many cases as sometimes the HIDL service takes even more than 500 ms to be ready.

So, just give the HIDL service more time, by restarting service after 1 second up to 10 times, for the total of ~10 seconds. This should be ample time for the HIDL service to initialize, and also reduce burst.

Note that StartLimitIntervalSec is specified at 20 instead of 10 to account for time it takes for bluebinder itself to attempt connection.

See systemd.unit(5) and systemd.service(5) for details on how these settings work.

Fixes: #3 (closed)

Merge request reports