Add PodDisruptionBudgets for multicluster components #7082
Comments
|
I'd like to take a stab at this 🙏 |
|
When starting to look into this I noticed there wasn't any tests (I could find at least) for the multicluster install command so I kicked this off with #7111 to get a baseline before changing any behaviour. |
adleong
pushed a commit
that referenced
this issue
Oct 20, 2021
Currently there are no tests of the output of the multicluster install command unlike the install and viz install commands. This makes it error prone to hard to validate changes. This is motivated by the addition of an ha mode to the multicluster components discussed in #7082. This change adds two test cases and refactors the install command to look like viz install making it easily testable. This means in practice that the body of the command is moved into an install function. Here we extract external data, eg. values, and delegates the values to a render function that handles the actual rendering. This is a non-functional change and the output used for the install_default.golden file is based of the main branch to validate this. Signed-off-by: Crevil <[email protected]>
|
Hey @Crevil it looks like to wrap this up we need to add high availability mode and a PDB to the service mirror. Any interest in wrapping this up? No problem if not—just thought I'd check first! |
olix0r
pushed a commit
that referenced
this issue
Apr 14, 2022
Currently there are no tests of the output of the multicluster install command unlike the install and viz install commands. This makes it error prone to hard to validate changes. This is motivated by the addition of an ha mode to the multicluster components discussed in #7082. This change adds two test cases and refactors the install command to look like viz install making it easily testable. This means in practice that the body of the command is moved into an install function. Here we extract external data, eg. values, and delegates the values to a render function that handles the actual rendering. This is a non-functional change and the output used for the install_default.golden file is based of the main branch to validate this. Signed-off-by: Crevil <[email protected]> (cherry picked from commit fddfb74) Signed-off-by: Oliver Gould <[email protected]>
This was referenced Jun 28, 2022
kleimkuhler
added a commit
that referenced
this issue
Jun 29, 2022
As part of #7082 we need to add HA mode to linkerd-multicluster's service mirror component. Before adding configuration for it, we should have a basic unit test that asserts the output of `linkerd multicluster link`. Signed-off-by: Kevin Leimkuhler <[email protected]>
|
Thanks for picking this up @kleimkuhler 🎉 |
|
Closing this as the remaining work is tracked by #8848. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Feature Request
What problem are you trying to solve?
When rolling the nodes of the cluster and/or scaling in it is possible for all the multicluster pods to be terminated.
How should the problem be solved?
The helm chart should provision
PodDisruptionBudgetsfor the multicluster components same as implemented in #5406 (discussed in #5398)What do you want to happen? Add any considered drawbacks.
Any alternatives you've considered?
Modyfing the output of the helm chart manually to add these features.
How would users interact with this feature?
They wouldn't as it should be a default for keeping the setup safe.
The text was updated successfully, but these errors were encountered: