Because old migrations haven't been removed from nova turbo-hipster
was happy to upgrade all the way to trunk from any old version.
However, we should step through each branch as that's the method
we as deployers to use.
Change-Id: Ie8d57379c316e25deef720bfa8872f38218dde1f
Python in ubuntu 12.04 needs an older version of greenlet. This is
a workaround while turbo-hipster is upgraded to 16.04.
Change-Id: Iaa1c806583f79378e40b76abf11bbce2a4361c45
To speed up bootstrapping time for turbo-hipster, check if the database
exists before creating it with the seed data as part of the job run.
This allows us to preload the database (as part of a nodepool image
build) which will save a lot of time.
Change-Id: Ieb09e714e24fe6df3ce148e44d5c1148739751c0
argparse was external in python 2.6 but not anymore, remove it from
requirements.
This should help with pip 8.0 that gets confused in this situation.
Installation of the external argparse is not needed.
Change-Id: Ib7e74912b36c1b5ccb514e31fac35efeff57378d
OpenStack projects are no longer being tested under Python 2.6, so
remove the trove classifier implying that this project supports 2.6.
Change-Id: I871e98dfbef99d8f50c47c435d21fb615e7e1eb9
The testing datasets are being updated to include this. This would
still be useful to test in the future if we fix kilo
Change-Id: I9d8849ce7b58eb5e2fd61c91ce691b58d27f3745
zuul_enqueue is a crude tool for rechecking turbo-hipster when
it goes haywire. There is no point rechecking patches that have
failed to merge.
Change-Id: Ib78736e5648131771b2872c30892b6beda6efc46
This patch exports a new env var for the script, the $TH_RESULT_FILE. If
the launched script writes something into the file name specified in
that env var and if the content of that file doesn't appear to indicate
a successful build, the content of the file is used as summary message
sent back to Zuul.
Originally, this patch attempted to do a similar thing through reading
back stuff from the shell's output log and reporting the last line from
that output, but the two major disadvantages were the inclusion of
timestamps in the log output, and Python's enrichment of the log line
with timestamps. I pondered writing a custom Python's logging.Handler
which simply remembers the last message, but the required changes in
utils.execute_to_log appeared rather messy to my untrained eye.
The main driver behind this change is a need to differentiate from hard
build failures (where we don't get any build artifacts because the build
simply failed) from "mere" test failures. Yes, our (KDE) tests are still
sloppy, and we have a fair amount of broken test still around, which is
why this difference matters to us. The result field in Zuul is more or
less free-form, so this patch does not impose any restrictions about its
format, except the "cannot start with 'SUCCESS'" because that one is
indeed very, very special.
It can be seen in action at http://ci-logs.kde.flaska.net/matrix.html .
Change-Id: I48c29d2566da12b02dcf27a551a058ecc4a303d4
The http://ci.openstack.org/ documentation site has been deprecated,
replaced by redirects to corresponding paths within
http://docs.openstack.org/infra/ where other Project Infrastructure
documentation already resides.
Change-Id: I8c7f0d87fafb8e6d2f41e4917008df0bbb603423
Rather than stepping through the migrations just in master (which
works fine assuming they are no different to previous releases)
checkout the correct stable branch and walk through that.
Change-Id: Iff79684990cbbaa8c6ed2474aca77a2f73d64b96
It's a little hacky but saves checking the logs for a failed upgrade
and then applying the migration and re-doing it. (Which would be
complicated due to the way handle_results works).
Change-Id: I8fd0c206c027f8cf984ba9fbd011cb6c6277f2b7
Fetching from git sometimes fails. Log the IP in these cases
so we can investigate the git.o.o logs
Change-Id: I09994cf0706670c44412f5d6da97c67329b40d1b
