Logins/Updates

[malins]

Hello,

I'm seeing dozens of logins/update operations per minute on my Keycloak instance originating from this provider, while I have only a couple of managed resources (1 client and 2 protocol mappers referencing this client). The managed resources are in state Ready=True/Synced=True, so I do not see any reason why constant logins/updates are required.

From they Keycloak logs:

2024-10-11 11:02:44,950 INFO  [org.keycloak.events] (executor-thread-87) type="LOGIN", realmId="03b5c1c7-e48a-498f-876c-cab6da425875", realmName="master", clientId="admin-cli", userId="504d731c-ef59-4c92-b0f8-80602250b63d", sessionId="6378022c-52f9-4a72-842e-c0cc80b92283", ipAddress="10.244.0.7", auth_method="openid-connect", token_id="9263a4f2-4af2-4270-a924-3ee960f49312", grant_type="password", refresh_token_type="Refresh", scope="email profile", refresh_token_id="24b45e81-9215-4701-a0e8-6eb0e327f8e3", client_auth_method="client-secret", username="admin", authSessionParentId="6378022c-52f9-4a72-842e-c0cc80b92283", authSessionTabId="VAL6LV5H_Yk"

2024-10-11 11:02:44,988 INFO  [org.keycloak.events] (executor-thread-87) operationType="UPDATE", realmId="03b5c1c7-e48a-498f-876c-cab6da425875", realmName="master", clientId="867589e4-a96b-4f6b-aab7-1b4cb95886d4", userId="504d731c-ef59-4c92-b0f8-80602250b63d", ipAddress="10.244.0.7", resourceType="CLIENT", resourcePath="clients/edf18451-8727-4a87-a0e4-77cea22a750c"

Is this a bug or is this normal behaviour? I'm running the latest version of this provider.

Thank you,
Manuel

[Breee]

That should be normal, as it is reconciling constantly to sync the desired state.
You can try to watch your ressources to see if they update really quickly with kubectl get <resource> -w , if yes then we have to check why.

maybe also related to #65 ?

[malins]

I was using a wrong (illegal) value for one of the fields of "Client". However, the update was accepted by Keycloak (and was corrected silently by Keycloak), and the provider discovered a "diff" just milliseconds after and therefore went into a very fast loop.

I'm not that deep into CP provider development, but throttling the update rate to a max of 1 update per second would be not a bad idea, altough this was my mistake at the end.