Merge lp:~javier.collado/utah/bug1169510 into lp:utah
Status: | Merged | ||||
---|---|---|---|---|---|
Approved by: | Javier Collado | ||||
Approved revision: | 876 | ||||
Merged at revision: | 866 | ||||
Proposed branch: | lp:~javier.collado/utah/bug1169510 | ||||
Merge into: | lp:utah | ||||
Diff against target: |
253 lines (+60/-54) 7 files modified
debian/changelog (+3/-0) examples/run_install_test.py (+0/-7) examples/run_utah_tests.py (+19/-19) utah/provisioning/provisioning.py (+1/-1) utah/provisioning/ssh.py (+24/-8) utah/run.py (+11/-16) utah/timeout.py (+2/-3) |
||||
To merge this branch: | bzr merge lp:~javier.collado/utah/bug1169510 | ||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Max Brustkern (community) | Approve | ||
Andy Doan (community) | Approve | ||
Review via email:
|
Description of the change
The changes in this branch are intended to make sure that timeouts are timely
handled when the utah client is running through SSH.
The problem here is that the SIGALRM isn't called until the call to the client
is finished (which might take hours) as explained in the following bug:
http://
To fix the problem, a timeout is set in the ssh socket, so that the call to the
client is interrupted in case of a timeout instead of waiting for the client to
finish before calling the signal handler.
In addition to this, there are a few exception related changes to make sure
that UTAHTimeout isn't caught by accident (making it a subclass of SystemExit)
and also to make sure that error messages are printed after cleanup messages to
make easier to troubleshoot problems.
LGTM - didn't know about SystemExit, so I've learned something new.