fix: a devpod up no longer outlives the dl that holds its lock - #668

Merged

blooop merged 19 commits into

mainfrom

fix/orphaned-devpod-up

Oct 2, 2026

Commits on Oct 2, 2026

fix: a devpod up no longer outlives a dl that dies without running its handler
blooop
committed
658ea6e
test: a SIGKILLed parent takes its own-group child with it, at the runner and at dl
blooop
committed
fb8220f
fix: rm and rme clear an orphan holding devpod's lock instead of naming kill
blooop
committed
2b26157
fix: the interrupt drain SIGKILLs a devpod up group that sits through its SIGTERM
blooop
committed
8363c45
docs: changelog entries for the orphaned devpod up fixes
blooop
committed
0da5263
test: a run whose SIGTERM was disarmed now loses its build to a Ctrl-C too
blooop
committed
ce93aeb
test: a Ctrl-C mid-up gives devpod its grace to unwind before the SIGKILL
blooop
committed
f7224e7
test: a SIGKILLed aid takes its boot down and the boot unlinks its token
blooop
committed
c203d31
docs: a blocked rm sweeps the lock rather than naming kill once
blooop
committed
de3f093
test: a delete that stays blocked on the lock sweeps it again
blooop
committed
fcb69d8
test: a later sweep that takes a once-spared holder is reported
blooop
committed
b88eacf
test: a delete blocked behind a live build signals nothing
blooop
committed
48c9a6a
release: 0.59.3
blooop
committed
6de25f9
fix: a blocked rm told the user kill ends a live build it cannot end
blooop
committed
03c59f8
test: the TERM row of the inherited-ignore table passed with the drain's SIGKILL gone
blooop
committed
40128b2
test: a --rm whose sweep freed the lock is told its own delete takes it
blooop
committed
0307b6b
test: a delete does not sweep its own blocked devpod delete
blooop
committed
33f88d6
docs: the drain's pid-reuse note credited the reap with what the live group guarantees
blooop
committed
bee0b5c
docs: kill's module doc still called the orphan's cause an open question
blooop
committed
5faba36

Navigation Menu

fix: a devpod up no longer outlives the dl that holds its lock - #668

fix: a devpod up no longer outlives the dl that holds its lock#668
blooop merged 19 commits into
mainfrom
fix/orphaned-devpod-up

Commits on Oct 2, 2026

fix: a devpod up no longer outlives a dl that dies without running its handler

test: a SIGKILLed parent takes its own-group child with it, at the runner and at dl

fix: rm and rme clear an orphan holding devpod's lock instead of naming kill

fix: the interrupt drain SIGKILLs a devpod up group that sits through its SIGTERM

docs: changelog entries for the orphaned devpod up fixes

test: a run whose SIGTERM was disarmed now loses its build to a Ctrl-C too

test: a Ctrl-C mid-up gives devpod its grace to unwind before the SIGKILL

test: a SIGKILLed aid takes its boot down and the boot unlinks its token

docs: a blocked rm sweeps the lock rather than naming kill once

test: a delete that stays blocked on the lock sweeps it again

test: a later sweep that takes a once-spared holder is reported

test: a delete blocked behind a live build signals nothing

release: 0.59.3

fix: a blocked rm told the user kill ends a live build it cannot end

test: the TERM row of the inherited-ignore table passed with the drain's SIGKILL gone

test: a --rm whose sweep freed the lock is told its own delete takes it

test: a delete does not sweep its own blocked devpod delete

docs: the drain's pid-reuse note credited the reap with what the live group guarantees

docs: kill's module doc still called the orphan's cause an open question