Skip to content

fix: remove duplicate proxy_read_timeout that crashes nginx on start #34

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
ronibhakta1 wants to merge 1 commit into from
Closed

fix: remove duplicate proxy_read_timeout that crashes nginx on start #34

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 2 additions & 5 deletions nginx/snippets/proxy-pass.conf
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,11 +4,8 @@
set $upstream_app app:8000;
proxy_pass http://$upstream_app;

# No backend synthesis timeout exists today (ffmpeg.py only bounds the
# encode step at 30s; TTS inference itself is unbounded) — pick a ceiling
# safely above realistic worst-case rather than nginx's low defaults cutting
# off legitimate slow requests.
proxy_read_timeout 120s;
# proxy_read_timeout is set per location (nginx rejects a directive repeated in
# one block, so it can't live here and be overridden after the include).
proxy_send_timeout 120s;
proxy_connect_timeout 5s;

Expand Down
9 changes: 6 additions & 3 deletions nginx/templates/20-speech-server.conf.template
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,9 +74,8 @@ server {
include /etc/nginx/snippets/proxy-pass.conf;

# TTS inference is CPU-bound and can be slow (24-layer models, long text,
# emulated arch) — give synthesis more than the shared 120s read timeout so
# a legitimately long generation isn't cut off with a 504. This override must
# come AFTER the include, whose 120s would otherwise win.
# emulated arch) — give synthesis more than the 120s default on other
# routes so a legitimately long generation isn't cut off with a 504.
proxy_read_timeout 300s;
}

Expand All @@ -91,5 +90,9 @@ server {
limit_conn perip 20;

include /etc/nginx/snippets/proxy-pass.conf;

# No backend timeout exists today (TTS inference is unbounded) — pick a
# ceiling safely above realistic worst-case rather than nginx's low default.
proxy_read_timeout 120s;
}
}
Loading