/v2/spaces/<spaceguid>/summary produces "-1" running_instances for all apps, instead of actual # instances

[metskem]

Issue

We upgraded to cf-deployment v24.1.0, which brings CAPI v1.142.0.
After the upgrade it appears that /v2/spaces/<spaceguid>/summary | jq .apps[].running_instances shows mostly “-1”.
Other calls that show the process states (like /v3/apps/<appguid>/processes/web/stats) do show the correct state "RUNNING".

Context

The issue is prominently shown when using the prometheus boshrelease, a CFAppUnHealthy alert fires for each app, since cf_exporter uses the /v2/spaces/<spaceguid>/summary a lot.

Steps to Reproduce

• Target any space that has one or more running apps.
• Find the space guid (cf space <spacename> --guid)
• Issue: cf curl /v2/spaces/<spaceguid>/summary | jq '.apps[].running_instances'

Expected result

The number of running instances for each app.

Current result

The result will be a "-1" for each app.

Possible Fix

I have no idea

[moleske]

cross referencing #3109 since seems like root cause is similar

[thelangley]

hey, just noticed that capi-release 1.143.0 was released yesterday which reverts the code which introduced this feature.

just tested in my env and we don't get -1 anymore

[metskem]

yup, same here, problem fixed.
Thanks!