Merge lp:~salgado/launchpad/kill-rogue-ec2-instances into lp:launchpad
Proposed by
Guilherme Salgado
on 2010-02-08
| Status: | Merged |
|---|---|
| Merged at revision: | not available |
| Proposed branch: | lp:~salgado/launchpad/kill-rogue-ec2-instances |
| Merge into: | lp:launchpad |
| Diff against target: |
74 lines (+15/-4) 3 files modified
lib/devscripts/ec2test/builtins.py (+2/-1) lib/devscripts/ec2test/instance.py (+1/-1) lib/devscripts/ec2test/testrunner.py (+12/-2) |
| To merge this branch: | bzr merge lp:~salgado/launchpad/kill-rogue-ec2-instances |
| Related bugs: |
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Jonathan Lange (community) | 2010-02-08 | Approve on 2010-02-08 | |
|
Review via email:
|
|||
To post a comment you must log in.
| Guilherme Salgado (salgado) wrote : | # |
| Jonathan Lange (jml) wrote : | # |
You're right, we need some sort of timeout, given that normal termination is unreliable. Ideally, the timeout would be based on inactivity, rather than total time, but I think this patch is a net win. Please land.
review:
Approve

= Summary =
Today, for the second time in less than 6 months, I started working and
realized an 'ec2 land' I ran (last Friday) on behalf of a community
member was hung and thus failed to bring the ec2 instance down.
This is pretty bad because in this case the instance was left up for
more than 48 hours, so I thought of adding scheduling a shutdown for 8h
after we start the preparations to run the test suite. This is more than
twice the time our test suite takes to complete, and it won't happen for
demo instances.