fix(helm): improve documentation and grafana dashboard

[genofire]

Pull Request Checklist

• I have added Go unit tests or Complement integration tests for this PR or I have justified why this PR doesn't need tests
• Pull request includes a sign off below using a legally identifiable name or I have already signed off privately

---

• docs: enabling of metrics in there config is needed
• dashboard: since the imported dashboard was created:
  • many metrics was dropped by dendrite (mainly Remove polylith/API mode #2967)
  • grafana has new version of diagrams ...

[genofire]

Again, something for you @S7evinK

https://github.com/genofire/dendrite/blob/patch-1/helm/dendrite/README.md

[genofire]

only possible metrics:

# HELP dendrite_caching_ristretto_cost 
# HELP dendrite_caching_ristretto_ratio 
# HELP dendrite_clientapi_reg_users_total Total number of registered users
# HELP dendrite_federationapi_destination_queues_backing_off 
# HELP dendrite_federationapi_destination_queues_running 
# HELP dendrite_federationapi_destination_queues_total 
# HELP dendrite_federationapi_recv_edus Number of incoming EDUs from remote servers
# HELP dendrite_syncapi_active_sync_requests The number of sync requests that are active right now
# HELP dendrite_syncapi_waiting_sync_requests The number of sync requests that are waiting to be woken by a notifier

[S7evinK]

Why are only those possible?

[genofire]

Dont ASK me, i believe the other was removed.

[S7evinK]

If I hit /metrics, I get these metrics as well:

# HELP dendrite_caching_ristretto_cost
# HELP dendrite_caching_ristretto_ratio
# HELP dendrite_clientapi_reg_users_total Total number of registered users
# HELP dendrite_clientapi_sendevent_duration_millis How long it takes to build and submit a new event from the client API to the roomserver
# HELP dendrite_federationapi_destination_queues_backing_off
# HELP dendrite_federationapi_destination_queues_running
# HELP dendrite_federationapi_destination_queues_total
# HELP dendrite_federationapi_recv_edus Number of incoming EDUs from remote servers
# HELP dendrite_federationapi_recv_pdus Number of incoming PDUs from remote servers with labels for success
# HELP dendrite_keyserver_device_list_update Number of times we have attempted to update device lists from this server
# HELP dendrite_roomserver_calculate_state_conflict_state_length The length of the conflicted room state.
# HELP dendrite_roomserver_calculate_state_duration_milliseconds How long it takes to calculate the state after a list of events
# HELP dendrite_roomserver_calculate_state_full_state_length The length of the full room state.
# HELP dendrite_roomserver_calculate_state_prev_event_length The length of the list of events to calculate the state after
# HELP dendrite_roomserver_input_backpressure How many events are queued for input for a given room
# HELP dendrite_roomserver_processroomevent_duration_millis How long it takes the roomserver to process an event
# HELP dendrite_syncapi_active_sync_requests The number of sync requests that are active right now
# HELP dendrite_syncapi_calculateHistoryVisibility_duration_millis How long it takes to calculate the history visibility
# HELP dendrite_syncapi_waiting_sync_requests The number of sync requests that are waiting to be woken by a notifier
# HELP download Total number of media_api requests for either thumbnails or full downloads
# HELP thumbnail Total number of media_api requests for either thumbnails or full downloads

The ones removed were those for internal requests, which were only useful for polylith deployments.

[genofire]

yes you are correct, after few incomming request that metrics are all visible ...

[genofire]

removed diagrams need metric:

• dendrite_caching_in_memory_lru_roomserver_room_ids (Room IDs)
• dendrite_caching_in_memory_lru_federation_event (Federation Events)
• dendrite_caching_in_memory_lru_server_key (Server Keys)
• dendrite_roomserver_processroomevent_duration_millis_bucket (Room Event Processing)
• dendrite_clientapi_sendevent_duration_millis_bucket (Sendevent Duration)
• process_cpu_seconds_total (CPU usage)