Our old check used an extremely inefficient boolean logic AND and
generated a bad query. This does a simpler check by querying the
inventories that the user has access to and using that as a simple
condition when performing the job template permission query
* upstream/release_2.2: (31 commits)
Updating changelog for 2.2 release
Bump version to 2.2.2
Use smart_str instead of .encode on passwords
Change default sensitive redaction behavior
Update changelogs for 2.2.1
Add an error message when running a job and trying to use an OpenSSH formatted key on an older version of OpenSSH.
Revert "Add sudo: true to the backup playbook"
Handle Ubuntu 12.04 psutil recursive cancel
fixed log viewer modal partial
fixed alignment of download standard out button in modals
About Tower version number fix
Pin package version for older release
add standard out download button to various places in the UI
Bump 2.2 branch to version 2.2.1
Don't create a group that is its own parent when an EC2 tag has an empty value. Fixes https://trello.com/c/2zc0odvX
Limit max depth when building mapping of group depths to avoid hitting recursion limit. Fixes https://trello.com/c/2zc0odvX
Fix .ini web links for v2
Add support for detecting encrypted openssh format private keys. Fixes https://trello.com/c/ZeVOXN5U
Fix psutil usage on el6 for job cancel
Set recursive on child process canceling
...
We originally tried to prune these .encode() calls out of the tree in
favor of smart_str but we must have missed this one. This allows
migration 0066 to work if utf-8 characters are used in the password
psutil will fail to import on el6 due to not being able to access a
pseudo terminal. This issues a SIGKILL to the proot process in order to
force the stop
Previously we could just check for the absence of the stdout file to
know we didn't yet/wouldn't ever have any stdout content. Now that we
are creating the stdout file ahead of time and persisting it we need to
also check to see if the file contains any data. This is also
necessary for some unit tests.
The act of committing a PR that including part of these changes and then
reverting it meant that some of the changes from the original branch
didn't land in this new PR. This commit adds the missing bits.
Lesson learned: Create a new branch when submitting a PR on a
previously reverted PR
* Cleanup old code related to stdout download using temporary files
* Remove copy-into-database code
* Modify stdout download code to reference stdout file
In addition to starting and stopping mongod in the correct order, the check for
determining whether mongod is required needs to happen when the service is
managed.
This is a temporary fix until we can get a permanent fix to the
performance issues we have with rendering this much data to the DOM. For
more info see my comment on https://trello.com/c/QrbeaZfK.
Fix a couple of typos to properly reflect package name and path.
Add vendored deps added when adding VMware & GCE support.
Add vendored deps added when vendoring pywinrm.
