Startup probe not working as expected about azure-container-apps HOT 5 OPEN

However, I am observing my replica being terminated after 3 attempts, i.e. 9 seconds.

@emilmuller the default values in the document only applies if you do not define the probe at all. If you define the probe while leave some property unset, it defaults to kubernetes defaults.
@anthonychu @cachai2 Probably we should clarify this in the doc to avoid confusion. Also, I spotted a the default timeout value for startup probe is 3 seconds. Moreover, I feel like it confuses customers by saying Failure threshold: timeoutSeconds. In fact the threshold does not related to timeout seconds if customer specifies them. So, we should just document the const default value. For StartupProbe it's 240 and for ReadinessProbe it's 48.

It would seem that groups of two health probes are executed almost at the same time. Are they failing, then immediately retrying and not respecting the 'period seconds' param?

@emilmuller could you send your container app name, environment name, resource group name, subscription id and region to acasupport#microsoft.com.

from azure-container-apps.

@zhenqxuMSFT Thanks. On a second, related note: Trying to figure out what the optimal initial delay seconds should be, and it's proving challenging. Looking through the logs, trying to sort on time to figure out when it starts.

There are four seeminly similar columns: TimeGenerated, time_t, _timestamp_d, TimeStamp_s. I can't find documentation on what the differences between these are, and which one you are suppose to sort on. Some of them yield millisecond variations, and you end up with different sorting. So far, I'm experiencing that sorting on _timestamp_d yields the most sensible sorting from what I can interpret from Reason_s. I'm ending up with this:

I'm guessing 'ContainerStarted' is the one I should be looking for to start measuring boot time.

It seems to log 'ImagePulled' twice. Not sure if something is happening twice when it only needs to happen once?

Regards

from azure-container-apps.

Hi @emilmuller, you are right, the log is duplicated and we have fixed this issue, it may take some time to deploy.
you can filter the log by time_t

from azure-container-apps.

This issue has been automatically marked as stale because it has been marked as requiring author feedback but has not had any activity for 4 days. It will be closed if no further activity occurs within 3 days of this comment.

from azure-container-apps.

This issue has been automatically marked as stale because it has been marked as requiring author feedback but has not had any activity for 4 days. It will be closed if no further activity occurs within 3 days of this comment.

from azure-container-apps.