support multiple replicas for linstor-controller #73
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
WanzenBug
force-pushed
the
controller-replicas
branch
2 times, most recently
from
2a9141e to
4c8a83a
Compare
|
We can get this merged now that we are no longer in an RC phase. Rebase needed, and the Helm keys need to be added to the new docs. |
WanzenBug
force-pushed
the
controller-replicas
branch
from
4c8a83a to
627cb7c
Compare
WanzenBug
force-pushed
the
controller-replicas
branch
from
627cb7c to
707a117
Compare
JoelColledge
previously approved these changes
Aug 7, 2020
|
I need to check if environment variables are passed to the ExecProbe. Then we could maybe find a cleaner solution |
WanzenBug
force-pushed
the
controller-replicas
branch
2 times, most recently
from
69802cb to
76b0b9f
Compare
|
So environment variables are available in ExecProbe During testing I noticed a few additional issues:
The second issue can be traced to the way we (ab-)use the |
WanzenBug
force-pushed
the
controller-replicas
branch
3 times, most recently
from
db6f6f0 to
066a083
Compare
JoelColledge
previously approved these changes
Aug 20, 2020
|
I want to polish our documentation a bit |
WanzenBug
force-pushed
the
controller-replicas
branch
from
066a083 to
092803b
Compare
WanzenBug
force-pushed
the
controller-replicas
branch
from
092803b to
2df767c
Compare
WanzenBug
force-pushed
the
controller-replicas
branch
from
2df767c to
67600ca
Compare
WanzenBug
force-pushed
the
controller-replicas
branch
from
67600ca to
02304a2
Compare
WanzenBug
force-pushed
the
controller-replicas
branch
from
02304a2 to
6c43999
Compare
Running multiple replicas requires special support from the linstor controller. The controller container will start a leader election process when it detects the presence of the K8S_AWAIT_ELECTION_* variables. The election process determines which pod is allowed to start the linstor-controller process. Only this pod will be added as endpoint for the controller service. Should the leader crash or the node its running on goes offline, a new leader will be elected and allowed to start the controller process. Note: in case the full node goes offline, the old pod will still be marked as ready. By using ClusterIP: "" on our service, we ensure we create an actual proxy (which automatically chooses the responding pod) instead of each client having to try multiple DNS responses.
WanzenBug
force-pushed
the
controller-replicas
branch
from
6c43999 to
1d526e1
Compare
|
@JoelColledge I think this is finally ready |
JoelColledge
approved these changes
Sep 25, 2020
|
🎉 |
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Running multiple replicas requires special support from the linstor
controller. The controller container will start a leader election
process when it detects the presence of the
K8S_AWAIT_ELECTION_*variables.
The election process determines which pod is allowed to start the
linstor-controller process. Only this pod will be marked as ready
and will receive traffic from the k8s service object.
Should the leader crash or the node its running on goes offline,
a new leader will be elected and allowed to start the controller process.
Note: in case the full node goes offline, the old pod will still be marked
as ready. By using
ClusterIP: ""on our service, we ensure we create an actualproxy (which automatically chooses a responding pod) instead of each client
having to deal with multiple DNS responses which may or may not respond.
See also: