Merge lp:~javier.collado/utah/install_sigterm_handler into lp:utah
Proposed by
Javier Collado
Status: | Merged | ||||||||
---|---|---|---|---|---|---|---|---|---|
Approved by: | Max Brustkern | ||||||||
Approved revision: | 819 | ||||||||
Merged at revision: | 819 | ||||||||
Proposed branch: | lp:~javier.collado/utah/install_sigterm_handler | ||||||||
Merge into: | lp:utah | ||||||||
Diff against target: |
70 lines (+25/-2) 3 files modified
examples/run_test_vm.py (+5/-1) examples/run_utah_tests.py (+1/-1) utah/run.py (+19/-0) |
||||||||
To merge this branch: | bzr merge lp:~javier.collado/utah/install_sigterm_handler | ||||||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Max Brustkern (community) | Approve | ||
Javier Collado (community) | Needs Resubmitting | ||
Review via email:
|
Description of the change
The changes in this branch add a handler for SIGTERM that just raises an
exception. This is intended to terminate the process gracefully and call all
the cleanup callbacks in the Machine object.
On one hand, I've verified that jenkins sends the SIGTERM signal when a job is
cancelled in the we interface. On the other hand, I've tested this
specificially with `run_test_
unmounted and the loop device freed when the SIGTERM signal is received.
To post a comment you must log in.
nice - all the changes to the run_*.py's are annoying. Could we do that from run.py safely, or is this best for now until we have a better way of managing these different script variations?