squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2107071044
While working on #5491 I realized there is a very important downside of this
way to detect a drift. We cannot change the calculation algorithm itself
without creating a disruption
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022422791
> > I know that a reconciliation is triggered but does the integration
filter apply ? because if so, then the previous and current integration object
would have been the same
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022290471
> I know that a reconciliation is triggered but does the integration filter
apply ? because if so, then the previous and current integration object would
have been the same (i.e.
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022262844
> is it true that if i.e. a deployment changes then we go to the same filter
as the Integration resource ? it seems to be odd
Yes. If you change a value of any Integration
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022272540
I know that an integration is triggered but does the integration filter
apply ? because if so, then the previous and current integration object would
have been the same, so the
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-209413
> > ```
> > * we can optionally compute the digest of the previous and current
object in the event filter (i.e. when the generation or phase have changed we
don't have to
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-200043
> * we can optionally compute the digest of the previous and current
object in the event filter (i.e. when the generation or phase have changed we
don't have to compute it)
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022156340
I wonder if we have to store them at all:
- the one for the pod is an ephemeral value that depends on the content of
an integration and some external resources, I don't think
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-2022101979
I had a look at this and the "problem" is that we use the digest as a mean
to compare the previous status of the Integration with the actual one in order
to trigger a reconciliation
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1969457828
The `CAMEL_K_DIGEST` env var was added to force the re-deploy of the pod in
case of a config map changes, like if only change the flow in an `Integration`,
then the pod won't
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1969431012
Okey. Indeed, it's because the env template vars in pod changes. I've
checked and it seems this var is not really used anywhere. I'll try to remove
it at all and run some test to
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1969218811
If you have an integration like this one (flow omitted because it does not
matter much):
```yaml
apiVersion: camel.apache.org/v1
kind: Integration
metadata:
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1968985697
Then, maybe this is not affecting the Pod execution at all. Do you have any
failing example, ie, when adding or removing a service without affecting
anything else?
--
This is an
lburgazzoli commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1968822133
Triggering a reconciliation is fine, but that should not lead to a
disruptive change in any resource so maybe we need two different digests or
rethink how the reconciliation is
squakez commented on issue #5192:
URL: https://github.com/apache/camel-k/issues/5192#issuecomment-1968672700
More than a bug I think it's the way it is actually designed. I'd mark this
as an enhancement. In any case I'm not sure how we should behave as most of the
traits configuration
15 matches
Mail list logo
