Overwrite the default command to pass discovery.type=single-node
Environment variable with "." can't be exported from bash.

See https://gitlab.com/gitlab-org/gitlab-runner/issues/2410

- Use flattened datatype (require elasticsearch 7.3)
- Upgrade to elasticsearch 7

JIRA INFRA-1295 #action In Progress

JIRA INFRA-1292 #action In Progress

JIRA INFRA-1291 #action In Progress

The core services update will trigger its own inventory sync on launch.
As the job takes some time to run, it avoids blocking the main inventory
sync. It also allows to trigger a core services update less frequently
by keeping only one in queue.

INFRA JIRA-1290

pre-commit assumes that the value of rev is an immutable ref (such as a
tag or SHA) and will cache based on that. Using a branch name (or HEAD)
for the value of rev is not supported and will only represent the state
of that mutable ref at the time of hook installation (and will NOT
update automatically).

Tell pytest to only look in the tests directory

New warnings after updating flake8 version used by pre-commit-hooks

A string or list is valid YAML but not valid for Ansible variables.
We expect only key-value-pairs (dict).

JIRA INFRA-1221

GitLab now accepts "." in the path

JIRA INFRA-1230 #action In Progress

Replace previous implementation based on the domain.

The domain can be changed per network but in practice we always use the
same domain in a network scope.
We currently have 9 network scopes but only 3 domains.
Using the network scope gives us finer granularity in practice.

JIRA INFRA-1228 #action In Progress

 '<' not supported between instances of 'NoneType' and 'int'

Sorting worked if all stack_members were None, but not if some were null
and some int.
Fixed by converting None to Infinity (so None values are at the end of
the list.

JIRA INFRA-1112 #action In Progress

Linked items are by default returned as a list of ICS id.
When recursive is True, a list of {ics_id, serial_number, stack_member}
is returned instead.

Items are always returned sorted by stack_member or ics_id when
stack_member is null.

Allow to add the serial_number and stack_member to the Ansible variables
in csentry dynamic inventory. Useful for switches.

JIRA INFRA-1111 #action In Progress

Default timeout increased to 30 minutes.
Some AWX tasks can be quite long especially as the csentry inventory
update can't run at the same time as other jobs.

JIRA INFRA-1051

Deferred tasks are now set to failed when the task it depends on fails.
No need to have an arbitrary time to disregard "old" deferred tasks.

JIRA INFRA-1051 #action In Progress

JIRA INFRA-1049 #action In Progress

The error has been raised a few times but the cause hasn't been found.
Adding the representation of the job should help debugging.

The caching of the load_user function already created several bugs
(INFRA 908/1018).
It requires to think about adding (or deleting in the admin view)
the current_user to the session. This is easy to miss.

It saves some database queries but we don't need this optimization.
Performances are not impacted and are good enough so far.

Removing the caching makes the application more robust and easier to
maintain.

JIRA INFRA-1025 #action In Progress

The LDAP_USER_DN was changed in 0.21.0.
This resulted in more entries returned when searching for users
and all users being disabled...

We were only looking into "OU=ESS Users" before.
Inactive users are put under "OU=InActiveUsers" in AD.

JIRA INFRA-1026 #action In Progress

LDAP_USER_DN was set to "OU=ESS Users" preventing to find service
account users.
Leaving it empty and relying only on LDAP_BASE_DN allows service users
to login.

JIRA INFRA-1022 #action In Progress

job.status shall be replaced with job.get_status() (RQ)

JIRA INFRA-1021 #action In Progress

- admin can delete any hosts
- normal users can delete hosts they created

Note that still only admin users can delete hosts via the API (more dangerous
operation as a script could delete many hosts).

JIRA INFRA-1018 #action In Progress

Regression from b9c1b838

Calling utils.trigger_ioc_repository_creation runs launch_task
which does a db.session.add(task). This raises:

SAWarning: Usage of the 'Session.add()' operation is not currently
supported within the execution stage of the flush process. Results may
not be consistent.  Consider using alternative event listeners or
connection-level operations instead.

SAWarning: Attribute history events accumulated on 1 previously clean
instances within inner-flush event handlers have been reset, and will
not result in database updates.

The result is that the host is not added to the elasticsearch index.

We can't trigger the task in after_update/after_insert events.
Perform the call in the view instead. It's temporary anyway. This should
be handled by IOC Factory.

JIRA INFRA-1015

The csentry_autoinstall_boot_profile variable shall be saved to setup the
DHCP option.
The DHCP shall be updated when a change occurs.

JIRA INFRA-1016 #action In Progress

The set event is triggered too early and no interface is yet associated
to the host. The fqdn is thus set to the name.

JIRA INFRA-1015 #action In Progress