Skip to content

Add keepalive messages to HTTP streaming connections#23355

Closed
komadori82 wants to merge 1 commit into
ggml-org:masterfrom
komadori82:keepalive
Closed

Add keepalive messages to HTTP streaming connections#23355
komadori82 wants to merge 1 commit into
ggml-org:masterfrom
komadori82:keepalive

Conversation

@komadori82
Copy link
Copy Markdown

Closes #22997

Long prompts can cause intermediaries to time out
and close idle connections after 300s (5m). This
patch adds a keepalive messaging system that lets
the intermediaries know we're still running during long waits.

Overview

Adds a keepalive to let http connectors know that we're not gone

Requirements

none

  • I have read and agree with the contributing guidelines
    yes
  • AI usage disclosure:
    yes, it helped me with adding the http keepalive code.

Closes ggml-org#22997

Long prompts can cause intermediaries to time out
and close idle connections after 300s (5m). This
patch adds a keepalive messaging system that lets
the intermediaries know we're still running during
long waits.
@komadori82 komadori82 marked this pull request as ready for review May 20, 2026 15:23
@komadori82 komadori82 requested a review from a team as a code owner May 20, 2026 15:23
@ngxson
Copy link
Copy Markdown
Contributor

ngxson commented May 21, 2026

the implementation is too hacky, we cannot accept. everything related to HTTP layer should stays at HTTP layer

@ngxson ngxson closed this May 21, 2026
@joleuger
Copy link
Copy Markdown

joleuger commented Jun 2, 2026

Similar change has now been merged: See #24013 and discussion in #23994

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Eval bug: stopping wait for next result due to should_stop condition when Prompt Processing is >60s

3 participants