Skip to content

feat: configurable alerts provider channel capacity#4364

Closed
siavashs wants to merge 1 commit intoprometheus:mainfrom
siavashs:feat-alert-channel-buffer-length-config
Closed

feat: configurable alerts provider channel capacity#4364
siavashs wants to merge 1 commit intoprometheus:mainfrom
siavashs:feat-alert-channel-buffer-length-config

Conversation

@siavashs
Copy link
Contributor

@siavashs siavashs commented Apr 23, 2025

This change makes the alerts memory provider channel capacity configurable.
The previously hard-coded value of 200 is used as default.
Bumping this value should result in better performance and less timeouts during sudden alert spikes, specially if the pipeline speed is affected by slow receivers.

The following new metrics are added with a subscriber label:

  • alertmanager_alerts_provider_channel_capacity
  • alertmanager_alerts_provider_channel_length

@siavashs siavashs force-pushed the feat-alert-channel-buffer-length-config branch 2 times, most recently from ced40e2 to 75bd989 Compare April 23, 2025 12:14
@grobinson-grafana
Copy link
Collaborator

Discussed here. I'm not against making this configurable, but I'm concerned that we don't understand enough about how and why this channel is becoming full and blocking. I've asked if we can get some profiles to better understand this.

@siavashs siavashs force-pushed the feat-alert-channel-buffer-length-config branch from 75bd989 to 5ae79c8 Compare June 18, 2025 11:04
@siavashs siavashs changed the title feat: configurable provider channel buffer length feat: configurable alerts provider channel capacity Jun 18, 2025
@siavashs siavashs force-pushed the feat-alert-channel-buffer-length-config branch 2 times, most recently from 548a3b5 to 6814d4f Compare June 18, 2025 11:31
This change makes the alerts memory provider channel capacity configurable.
The previously hard-coded value of 200 is used as default.
Bumping this value should result in better performance and less timeouts during sudden alert spikes,
specially if the pipeline speed is affected by slow receivers.

The following new metrics are added with a `subscriber` label:
- alertmanager_alerts_provider_channel_capacity
- alertmanager_alerts_provider_channel_length

Signed-off-by: Siavash Safi <siavash@cloudflare.com>
@siavashs
Copy link
Contributor Author

Closing this, see the comment is linked issue.

@siavashs siavashs closed this Oct 13, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants