Releases: bww/psctl
v0.8.2
v0.8.0
Adds support for defining per-process environment variables in taskfiles.
version: 1
tasks:
-
name: a
env:
MESSAGE: All your base are belong to us.
run: echo "$MESSAGE"v0.7.1
Adding short versions of some command line flags. Refer to psctl -h.
v0.7.0
Improves managed process output. STDOUT and STDERR is now prefixed with it's label and, when applicable, colorized.
v0.6.0
Adds support for a new kind of availability check: shell:<command>. A shell check runs an arbitrary command and is considered to have passed when the exit status of that command is 0.
version: 1
tasks:
-
name: a
run: ./service_a
check:
- shell:nc -z localhost 9876
-
name: b
run: ./service_b
check:
- http://localhost:8080/statusThis version also fixes an issue where a process that has already exited will continue to poll for availability checks which cannot possibly succeed.
v0.5.0-1
Adding support for configurable wait duration in a process specfile:
version: 1
tasks:
-
name: tests
run: ./test_harness
deps:
- api
-
name: api
run: build/api
wait: 60s
checks:
- http://localhost:4000/status
v0.4.0
Further improves cleanup of processes. Specifically, processes are now cleaned up after PSCTL is interrupted.
v0.3.3
Improves cleanup of managed processes when PSCTL shuts down because a process with dependency checks never became available before its deadline. Previously, PSCTL would shut down without explicitly cleaning up after the processes it had started to that point. Now it explicitly cleans up any processes that had been started.
v0.3.2
This release actually improves cleanup for child processes that are killed by Process Control because the system is coming down.
Previously, on some Unix systems, these terminated processes could become zombies and subprocesses of those processes would never be cleaned up. Processes are now managed via process groups, which are now explicitly killed to reclaim their resources and avoid this state.
v0.3.0
This release improves cleanup for child processes that are killed by Process Control because the system is coming down. Previously, on some Unix systems, these terminated processes could become zombies. They are now explicitly reaped after being killed to reclaim their resources and avoid this state.