| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
The current link does no longer work.
|
|
this supports a list of input, and cleans up that list
for the platform specific mount types. Basically,
mtype = None
means 'mount -t auto' or the equivalent for the platform.
and 'iso9660' means "iso type".
|
|
|
|
This allows the caller to supply their own leaf decoder, which will
then be in charge of translating the content of the url.
|
|
HVM instances on EC2 have grub on /dev/xvda.
The bug here resulted in a prompt on grub update.
LP: #1336855
|
|
add kwargs to fork_cb, and utilize that to call log_time and pass through
the provided args to resize_cmd.
LP: #1338614
|
|
|
|
|
|
util.log_time()'s return value was what was being sent to fork_cb. This means
the resize ran in parallel and the call to fork_cb threw a traceback (trying
to call Nonetype).
By permitting fork_cb to take kwargs, and using the correct method syntax,
this now forks and resizes in the background as appropriate.
|
|
|
|
silently wait 5 seconds for networking to come up.
We started seeing the message more now, as we are now blocking
the networking from coming up until cloud-init-local is done.
Previously that would have happened in parallel, and we were less likely
to see that message.
|
|
This makes it so networking wont start to come up until after
cloud-init-local has had a chance to search local datasources
and write /etc/network/interfaces.
The changes most likely need to still be done for systemd.
LP: #1368861
|
|
not sure why this was here.
the intent must of have been to allow for a local datasource
to continue booting and not annoyingly block waiting for network information
(if ithere was no network information).
However, that seems wrong.
If the datasource wipes /etc/network/interfaces and there are no network
interfaces then we're probably breaking that use case here. However we're
fixing the other more common case.
|
|
|
|
Not all vendor data is destined for cloud-init. This sanely reads
the vendor data as a dict, array or a string.
|
|
|
|
This makes it so networking wont start to come up until after
cloud-init-local has had a chance to search local datasources
and set /etc/network/interfaces.
The changes most likely need to still be done for systemd.
LP: #1368861
|
|
|
|
With the writing of cloud-init status, cloud-init-local needs to
have /run mounted. The issue we were seeing was a race where:
cloud-init-local creates /run/cloud-init
/run is mounted
cloud-init-local tries to link a file into /run/cloud-init
that directory was no longer visisable as /run was mounted over
the top.
LP: #1353008
|
|
|
|
For now, this vendor data handling is just added to openstack.
However, in an effort to allow sanely handling of multi-part vendor-data
that is namespaced, we add openstack.convert_vendordata_json .
That basically takes whatever was loaded from vendordata and takes
the 'cloud-init' key if it is a dict. This way the author can
namespace cloud-init, basically telling it to ignore everything else.
|
|
We were checking for presense of meta_data.json for each supported
metadata version. Instead just check that /openstack is there.
This reduces the time to check on EC2 or any other cloud.
|
|
This makes the DataSourceConfigDrive support vendor-data in the same
way the metadata service reader does. There are still some things to
fix here, but now we're similar between these two.
Also drops the ability to specify a version (as in YYYY-MM-DD) that you want to
look for. Nothing was using this, but it may be useful to add back in
in the future and expose as a datasource config option.
|
|
instead of taking a version that they should look for,
the readers now just select the highest supported version.
definitely a use case later for having version= but nothing
is using it now.
|
|
|
|
|
|
|
|
I had leaked in a couple exception loud-ness changes previously.
This does fix one bug in that os_versions was not defined.
|
|
using tuple for _versions was just not necessary.
fix reference to undefined os_versions.
|
|
If something is broken as in a built in config, or code
just broken, then logging warning during search for metadata
is ok.
|
|
|
|
make pyflakes now passes.
|
|
This data will be treated the same as vendordata from other sources.
|
|
|
|
Updated read_config_drive: removed the unused version kwarg, used the
OS_VERSIONS tuple from the openstack helper to avoid hardcoding
versions.
Added a comment to the tuple in helpers/openstack.py asking for it to
be kept in chronological order.
|
|
In a container the device nodes may exist but not be writable.
I'm seeing this on trusty host with trusty containers, the root
device ends up looking like it is to /dev/loop0.
LP: #1366891
|
|
Use the regular logic to create sudo rules and just supply the correct
filename. The temp script in tools/ should install 2 more dependencies.
|
|
|
|
|
|
|
|
|
|
(with pkg).
|
|
|
|
|
|
|
|
Instead of using this log (which really isn't a failure) we should
instead of just return the looked up locations and then if there really
is an error the caller can handle the usage of the looked up locations
as they choose fit.
|
