You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Problem: when using procd init scripts, the started service cannot be stopped, when the respawning is in progress.
The test case for the simple start/stop case is attached. Save attached shell scripts procd-test in /etc/init.d, run-test in / and run the test - the /run-test.log output is attached too.
Same can be observed in more complicated test scripts too, for example removing a respawning service instance is not possible (switch from one instance to no instance - procd should just watch the registered configuration/interfaces for changes and restart when change happens) - the instance is still listed in "ubus call service list" and is actively respawning.
This is a problem, because when for example the service is running on PPPoE interface, which goes down for a minute, the service fails and is respawned several times by procd even when the init script tried to remove the instance. As a multi-fail consequence, the procd removes the respawn parameters and does not reintroduce them when the PPPoE is back up and the init script starts the instance again. Result is a started service, but with no respawning.
Expected result: procd should stop respawning as soon as possible and apply the requested changes afterwards.
Tested on OpenWrt git master (from github) on TP-Link TL-WR1043ND and on x86/64bit Hyper-V virtual machine (VDI converted to VHDX), both give the same result.
The text was updated successfully, but these errors were encountered:
oldium:
Problem: when using procd init scripts, the started service cannot be stopped, when the respawning is in progress.
The test case for the simple start/stop case is attached. Save attached shell scripts procd-test in /etc/init.d, run-test in / and run the test - the /run-test.log output is attached too.
Same can be observed in more complicated test scripts too, for example removing a respawning service instance is not possible (switch from one instance to no instance - procd should just watch the registered configuration/interfaces for changes and restart when change happens) - the instance is still listed in "ubus call service list" and is actively respawning.
This is a problem, because when for example the service is running on PPPoE interface, which goes down for a minute, the service fails and is respawned several times by procd even when the init script tried to remove the instance. As a multi-fail consequence, the procd removes the respawn parameters and does not reintroduce them when the PPPoE is back up and the init script starts the instance again. Result is a started service, but with no respawning.
Expected result: procd should stop respawning as soon as possible and apply the requested changes afterwards.
Tested on OpenWrt git master (from github) on TP-Link TL-WR1043ND and on x86/64bit Hyper-V virtual machine (VDI converted to VHDX), both give the same result.
The text was updated successfully, but these errors were encountered: