Fix/etcd data validation #93
Conversation
georgekuruvillak
requested review from
shreyas-s-rao and
swapnilgm
as code owners
georgekuruvillak
force-pushed
the
fix/etcd-data-validation
branch
3 times, most recently
from
a0be0da to
f39faa6
Compare
georgekuruvillak
added
component/etcd-backup-restore
There was a problem hiding this comment.
@georgekuruvillak Thanks for the well-written PR! Just a few typos and suggestions. Other than that, LGTM! 👍🏼
|
@georgekuruvillak Please change the bootstrap script in the helm chart as well, at Also, in our example YAMLs, the label selector in the Service spec (app: etcd) does not match the labels given for the StatefulSet (app: etcd-${cloud}) at Could you please make the necessary changes for this as well? Thanks. |
georgekuruvillak
changed the title
doc/validation.md
Outdated
| * Do directory content validation. | ||
| * Start etcd | ||
| * Yes | ||
| * Check if previous exit was normal |
There was a problem hiding this comment.
Is this second check required?
There was a problem hiding this comment.
No. I have changed the proposal as how the validation is done now.
| @@ -23,23 +23,102 @@ metadata: | |||
| data: | |||
| bootstrap.sh: |- | |||
There was a problem hiding this comment.
Does the bootstrap.sh be specific to the cloud provider?
There was a problem hiding this comment.
It is generated from the template file. So the code generator generates across all the cloud providers. This will be changed when the helm charts are introduced.
example/etcd-statefulset-aws.yaml
Outdated
| done | ||
| } | ||
|
|
||
| if [ ! -f $MARKER ] ; |
There was a problem hiding this comment.
| if [ ! -f $MARKER ] ; | |
| if [ ! -f $VALIDATION_MARKER ] ; |
georgekuruvillak
force-pushed
the
fix/etcd-data-validation
branch
2 times, most recently
from
e82b5f1 to
02e6e14
Compare
|
@georgekuruvillak Can you also add a test or two for testing in sanity mode? And also modify the tests to pass the mode argument to the |
georgekuruvillak
force-pushed
the
fix/etcd-data-validation
branch
from
02e6e14 to
d4a0a5f
Compare
I have made the changes and have added test for sanity check. PTAL |
There was a problem hiding this comment.
@georgekuruvillak Thanks for making all the suggested changes. LGTM! 👍🏼
shreyas-s-rao
added
reviewed/ok-to-test
gardener-robot-ci-1
added
needs/ok-to-test
What this PR does / why we need it:
Design decision on how to go ahead with data validation.
Which issue(s) this PR fixes:
Fixes #71
Etcd data directory validation takes a long time and resource to perform. Currently, it gets triggered when etcd restarts. This PR changes the logic to perform full validation only when etcd restarts erroneously.
Special notes for your reviewer:
Please go through the design doc so that we can come to a consensus on how to proceed ahead with the validation steps and division of responsibilities between the script in etcd container and the sidecar.
Release note: