agent-state: error details: hook failed: "install"
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Ubuntu CI Engine |
Fix Released
|
Critical
|
Vincent Ladeuil | ||
Ubuntu CI Services |
Fix Released
|
Critical
|
Vincent Ladeuil |
Bug Description
For the third time I end up with a deployment failing with:
unit: lander-jenkins/0: machine: 19 agent-state: error details: hook failed: "config-changed"
unit: bsb-restish/0: machine: 15 agent-state: error details: hook failed: "install"
Looking at /var/log/
2014-03-05 18:39:09 INFO juju-log grabbing service from tarball...
2014-03-05 18:39:09 INFO install % Total % Received % Xferd Average Speed
Time Time Time Current
2014-03-05 18:39:09 INFO install Dload Upload
Total Spent Left Speed
13 59.3M 13 8448k 0 0 98674 0 0:16:44 0:01:26 0:15:18 61968
13 59.3M 13 8448k 0 0 94459 0 0:10:52 0:01:30 0:09:22 77932
13 59.3M 13 8448k 0 0 94459 0 0:10:59 0:01:31 0:09:28 0
2014-03-05 18:40:45 INFO install curl: (18) transfer closed with 53615230 bytes
remaining to read
2014-03-05 18:40:45 INFO install
2014-03-05 18:40:45 INFO install gzip: stdin: unexpected end of file
2014-03-05 18:40:45 INFO install tar: Unexpected EOF in archive
2014-03-05 18:40:45 INFO install tar: Unexpected EOF in archive
2014-03-05 18:40:45 INFO install tar: Error is not recoverable: exiting now
2014-03-05 18:40:46 INFO juju-log Error running: curl https:/
canonical.
tar -xzC /srv/bsb_restish: None
2014-03-05 18:40:46 ERROR juju.worker.uniter uniter.go:480 hook failed: exit sta
tus 2
2014-03-06 10:54:24 INFO juju-log Running hook: config-changed
2014-03-06 10:54:25 INFO juju-log Running hook: start
At first I thought I had a corrupted payload due to some mistake on my part (like two concurrent swift uploads or something), but that last occurrence ruled out that scenario.
I 'swift download ci-payload'ed and indeed it's corrupted >-/
Why it breaks only some unit installs is a mystery...
Related branches
- Evan (community): Approve
- PS Jenkins bot (community): Approve (continuous-integration)
-
Diff: 427 lines (+145/-89)12 files modified.bzrignore (+2/-0)
branch-source-builder/bsbuilder/resources/v1.py (+1/-1)
branch-source-builder/bsbuilder/tests/test_v1.py (+1/-1)
branches/README (+6/-0)
image-builder/imagebuilder/resources/v1.py (+1/-1)
juju-deployer/deploy.py (+89/-67)
juju-deployer/test_deploy.py (+7/-3)
ppa-assigner/ppa_assigner/api.py (+1/-1)
tests/run.py (+28/-10)
tests/test_run.py (+7/-3)
ticket_system/ticket/api.py (+1/-1)
ticket_system/ticket_system/urls.py (+1/-1)
Changed in ubuntu-ci-services-itself: | |
status: | In Progress → Fix Committed |
Changed in ubuntu-ci-services-itself: | |
status: | Fix Committed → In Progress |
Changed in ubuntu-ci-services-itself: | |
status: | In Progress → Fix Committed |
Changed in ubuntu-ci-services-itself: | |
status: | Fix Committed → Fix Released |
tags: | added: airline |
Changed in uci-engine: | |
assignee: | nobody → Vincent Ladeuil (vila) |
importance: | Undecided → Critical |
milestone: | none → phase-0 |
status: | New → Fix Released |
I'm working on a fix to reduce the payload size.
I investigated further the failure above and ended up with the feeling that using a StringIO to hold a >60M file has some issues, as well as loading that in memory to upload to swift.
swift put_object accept a file handle (or any object that provides read()) instead of a string, I'm using that in my fix.