A few Markdown-oriented issues were causing Docutils errors to
leak into the end-user docs on http://devstack.org
Change-Id: I51fa9698afb1bfb48596478d83bd1fdcd84ac52e
tools/build_docs.sh generates the devstack.org website from the static
pages and generated pages created by running shocco against a DevStack
checkout.
Note that while this is the complete auto page generation of the
devstack.org site, pushing the content back to GitHub is limited
to those with push access to the current repo.
Partial-bug 1235626
Change-Id: I61dc3d56e4a4832a9ddd1904dd8af65c15a17e50
When 'keystone token-get' fails, the caller can't detect the failure.
This cause troulbe shooting a bit complicated.
Change-Id: I3c58c5fd0e92a87e87546ea797904e08646a1097
Closes-Bug: #1238412
unlike our python code, we have no automatic style checking for
bash. For the most part, it's not a big deal, but errant whitespace
or incorrect indenting is sometimes annoying to have to -1 people's
patches for. Instead of constantly picking it up in manual review
maybe we can do better.
This is an uber hacky script which could be used to do just that.
./tools/bash8.py file1 file2 file3 ...
And it will show issues found with the files at hand. Lightly
tested in the existing devstack tree, it exposes a few issues that
we might want to think about.
This should be python 3 compatible, and includes argparse to provide
a basic '-h' support to explain how the command should be run.
Change-Id: I5009fa5852595c2953a548e430e5e1ce06ae94e0
-The change in install_docker.sh corrects a typo ('=' should be '-').
This typo resutls in 'unable to locate the packet' error when executing apt-get.
-The second change is in hypervisor-docker fix the error for reporting
docker is not installed when docker is actually set up. The original line
missed the version part of the package name.
Change-Id: Ic48f45158cf84f89080f095d53c355e9f6969bfd
Run ./stack.sh will dump ~400 lines of information, because of
tar xvfz pip-*.tar.gz, and python setup.py install.
We'd better mute stdout for the two steps, to make console cleaner
Change-Id: Icf87947e020acb48d8cbe4cdcc1641f060e50f6d
'set -o errexit' recently added to the pip installer script, which causes
the script fail when it does not able to find an already installed pip.
This change handles the situation when pip is not installed.
Change-Id: I18a42d13c4be6699db21ec5b6a095a88a199912d
Automatically creating a new user account is not always the right course
of action when stack.sh is running as root. Plus, the re-exec did not
work correctly in some cases.
* Create tools/create-stack-user.sh to set up a suitable user
for running DevStack
* Abort stack.sh and unstack.sh if running as root and suggest creating a
suitable user account.
Change-Id: I5d967c00c89f32e861449234ea8fe19261cd9ae3
get_packages() always included 'general' as a default 'service' file.
Remove this assumption and add it explicitly to the primary package
installation call. This allows get_package() to be used in other places
where 'general' is not desired to be included.
Change-Id: I1eed4386d073d6ae9534aedae32654208c6662e8
Give as much VCPUs to the DevStack machine as possible. First asking
xenapi about its CPU count, and as a fallback, count the CPUs in dom0.
This should result in faster test runs.
Change-Id: I1ffb99ecd435f1d7eb5754fe9cd99f0e8ceae6dc
stack.sh invokes some helper scripts as separate processes, rather than
by source'ing them. As with stack.sh itself, abort immediately on the
first error, so that errors don't compound and result in confusing error
messages. If one of these helper scripts aborts, stack.sh itself will
also abort in the usual manner.
Due to the change in behaviour, tweak some mv invocations to ensure that
they don't trigger false failures.
As with stack.sh itself, also enable xtrace so we can see exactly what's
happening. In particular this allows us to see the cause of any
premature termination due to a command failing whilst errexit is
enabled.
Change-Id: I7a55784c31e5395e29ab9bbe2bb112b83b9be693
xenbr0 is correct for most installations, but not all. Notable xenserver-core
may use a differently named device. Since we can auto detect this, remove the
config and do so.
Change-Id: I989f6ddd5ffb526ab350f263ef6fc402c596304a
Set the DHCP timeout to 120 seconds during virtual machine installation.
Some users failed to run devstack, due to a low DHCP timeout setting.
The default value is 60 seconds. This change sets the value to 120 secs,
that should give enough time for most people.
Change-Id: I15fde45ed0d005c1a8621134eee6c3c338b5be5d
By exporting the LC_ALL=C we can get rid of localisation issues, as the
actual scripts are already assuming an english installation.
FIxes bug 1227527
Change-Id: Ieeebce4d53b09959146a970f3fb803201ac5ebdf
install_docker.sh failed for me because socat wasn't installed. Add it
to this script since it expects it to be there.
Change-Id: Ic55f5e38de1b38bdd37407b7bec533d4c3eff2a9
This change updates the readme, so that it is easier to get started, and
reflects the actual behavior of DevStack.
Changes in README.md:
- Link to xenserver download page
- Remove neutron interface - it is no longer installed by devstack
- Add appendix with
- How to use a different ubuntu mirror
- How to use a proxy for ubuntu
- How to re-use the created VM
- Remove run from snapshot section and "do cloudy stuff"
Changes in the Readme-embedded sample localrc:
- Upload a vhd image and a uec image by default - easier to get started
Change-Id: I13bb8e59ff5367ff7623fe9aa273886a957f81a7
In XenServer scenarios, an additional domU is created to run OpenStack
services. This change is increasing the memory for that VM to speed up
test runs.
Change-Id: I322f4e4703e506620fa7e7456c4264ee0d050edc