Reduce the prod-playbooks #379
Conversation
| - name: ome.omero_web | ||
| version: 4.0.1 | ||
|
|
||
| - src: ome.java |
There was a problem hiding this comment.
Are these expected to be examples or the actual playbooks used for deploying the production services?
In the latter case, I think you might want to keep the version rather than installing the latest role by default
There was a problem hiding this comment.
Pin the versions of roles:
The drawback of this is that the versions will become stale (that happened already).
The advantage of this - I am not sure about the disadvantages implied, maybe we should discuss this ? I thought our roles are pretty well maintained and we do not want to introduce anything to them which is not compliant with our prod playbooks deployment ?
There was a problem hiding this comment.
The drawback of this is that the versions will become stale (that happened already).
That's correct, if these are stored, these versions need to be maintained, that's not different from any other libraries with dependencies.
The advantage of this - I am not sure, maybe we should discuss this ?
Primarily a reproducibility argument. If a new version of the role with backwards-incompatible changes is made, this approach means installing the roles and running the playbook from the exact same commit could result in a different outcome.
|
|
||
| At the moment, only the `ome-demoserver.yml` playbook is available here. This is a playbook for OMERO.demo server on https://demo.openmicroscopy.org OMERO.server and OMERO.web. You can read more about this [OMERO.demo server on our website](https://www.openmicroscopy.org/explore/). | ||
|
|
||
| We are in the process of adding more OME team's production playbooks here. |
There was a problem hiding this comment.
Is the intent of this repository to only hold OMERO production playbooks moving forward?
There was a problem hiding this comment.
Is the intent of this repository to only hold OMERO production playbooks moving forward?
That was my initial intent when planning this PR and discussing the tactics previously with @jburel .
What else would you like to hold here @sbesson ?
The idea is to minimize and simplify wherever possible.
But happy to discuss about better strategies.
There was a problem hiding this comment.
I don't have anything different to propose. I was mostly enquiring as I saw the omero subfolder was maintained and this was historically used for separating the OMERO playbooks from the others.
There was a problem hiding this comment.
Yes, omero folder is maintained to get out of the conandrum of molecule not finding the roles folder. Maybe I could rename the folder to playbooks ?
pwalczysko
force-pushed
the
only-demo-on-prod-playbooks
branch
from
ecb6a70 to
c1461fe
Compare
|
👍 For the demo server email changes! |
|
The fixes were pushed, the new changes are implemented too. |
There was a problem hiding this comment.
Review is quite tricky here as the scope of the changes is quite big (>4K lines deleted, complete overhaul). Focusing exclusively on the diff for the non-deleted files, the changes to the playbook and the template files are fairly minimal. The requirements files have been reordered and there is no easy way to review the version bump but the Molecule scenarios are upgraded to run on Rocky Linux 9 and passing.
Assuming this PR represents the state of the current deployment of the various OMERO production servers at UoD, my recommendation would certainly be to merge this sooner rather than later.
|
@jburel Are you happy for me to merge this please ? |
.github/workflows/molecule.yml
Outdated
| strategy: | ||
| # Keep running so we can see if other tests pass | ||
| fail-fast: false | ||
| matrix: | ||
| scenario: ${{fromJson(needs.list-scenarios.outputs.matrix)}} | ||
| steps: | ||
| - uses: actions/checkout@v3 | ||
| - uses: actions/checkout@v4 | ||
| - uses: actions/setup-python@v4 |
|
Merging as discussed at the meeting. thanks All. |
This PR is replacing #378
The idea is to slim down the prod-playbooks repo to a more manageable size of 1-2 playbooks which are offered to community as examples.
The plan (to be executed in 2 stages) is to
ome-demoserver.ymlplaybook and test it via moleculeThis PR is tackling step 1. above:
ome-demoserver.ymlome-demoserver.ymlplaybook for Rocky Linux 9 / RHEL 9 as it is used now in our private repo in productionsite.ymllogic as it is too complex for just 1-2 playbooks and does not help the users when attempting to run/use this repo imhotemplateswith the private repo ones -> this involves update ofomero-web-config-for-webapps.j2(tagsearch and webtagging)create,converge..., i.e. the install of OMERO.server and OMERO.web as perome-demoserver.ymlplaybook) to themolecule/ome-demoserver/molecule.yml- create a
prepare.ymlfor install ofcron- add the
inventoryparagraph tomolecule.ymlto pass the correct variable values to the testing framework- make sure the roles during the test are installed where the molecule machinery can find them (change in
ansible.cfg)- run the dependency step before the first destroy step, as otherwise the destroy step crashes
I have also verified that the playbook
ome-demoserver.ymlis working fine on the production demo server in check mode when run from this repo exactly as prescribed now in the README cmds.cc @khaledk2 @sbesson @jburel
-