Repository navigation
server: use port 9931 by default - #30159
Merged
Merged
Conversation
ngxson
requested review from
a team,
CISC,
danbev and
ggerganov
as code owners
October 8, 2026 13:43
github-actions
Bot
added
documentation
ServeurpersoCom
approved these changes
Oct 8, 2026
Contributor
|
@ngxson the port change looks complete to me, but the commit also carries unrelated changes in ggml/src/ggml-cpu/ggml-cpu.c and src/llama-graph.cpp (the graph plan skip and the build_inp_embd reorder), probably leaked from another branch: could you drop them from this PR? |
CISC
approved these changes
Oct 9, 2026
CISC
added
the
highlight
feal87
added a commit
to feal87/myllama.cpp
that referenced
this pull request
Oct 9, 2026
Upstream brings the chat API refactor into common_chat_session (PR ggml-org#30210), the UI models manager (ggml-org#29583), server default port 9931 (ggml-org#30159), and CUDA / metal / sycl / vulkan / musa fixes. Conflicts and resolution: - server-context.cpp: upstream moved message spans into task.apply_chat_session() and dropped the per-request message_delimiters JSON round-trip. Keep the fork deferred-mtmd loop bound (n_inputs) and pass chat_session.message_delimiters() to task.mtmd_delims. - server-queue.cpp: keep the fork reported_tool_names / base_tool_names / markdown_fences bookkeeping and adapt it to upstream's task_result_state() free constructor. Kept the fork MoE expert cache (src/llama-moecache.*); scripts/fork-post-merge.sh reports no upstream MoE cache leftovers. Assisted-by: pi (deepseek-flash)
6 of 7 tasks
liu-x27
added a commit
to liu-x27/XavierJev
that referenced
this pull request
Oct 18, 2026
The quickstart started llama-server without a port and then set AGENT_JUDGE_BASE_URL to http://127.0.0.1:8080/v1, which was llama-server's default. llama.cpp v0.7.0 (10-15, ggml-org/llama.cpp#30159) moved the default to 9931, so on a current build the two lines no longer meet: the judge cannot be reached and the startup self-check refuses to start, safe but stuck. The command now says --port 8080. eval/llama-cpp/run.ts already passes its own port. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
xiaoyu607
pushed a commit
to xiaoyu607/reika
that referenced
this pull request
Oct 22, 2026
llama.cpp v0.7.0 moved llama-server's default port from 8080 to 9931 (ggml-org/llama.cpp#30159); its Docker images keep 8080 inside the container. The fallback exists so a fresh install works with no config, so it follows upstream. A refused connection at the default now names 8080 and REIKA_BASE_URL instead of a bare "fetch failed", for anyone on an older llama-server or the Docker image. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Overview
Ref #26116 #26508
For docker, we still use 8080 inside the container to avoid breaking any tools using our base image
Requirements