Skip to content

Fix/caddy ws flush and upgrade limit - #1

Merged
nol4lej merged 3 commits into
mainfrom
fix/caddy-ws-flush-and-upgrade-limit
Jul 28, 2026
Merged

Fix/caddy ws flush and upgrade limit#1
nol4lej merged 3 commits into
mainfrom
fix/caddy-ws-flush-and-upgrade-limit

Conversation

@nol4lej

@nol4lej nol4lej commented Jul 28, 2026

Copy link
Copy Markdown
Member

No description provided.

nol4lej added 3 commits July 26, 2026 23:41
Disable response buffering on the RPC upstream (flush_interval -1) so
WebSocket frames pass through immediately. Buffering stalled long-lived
subscription sockets, which surfaced as WS connections dying after a few
seconds and clients reconnecting in a loop.

Raise the WS upgrade limit from 10 to 30 events / 60s on both the
CF-Connecting-IP and direct-origin zones, so the reconnect storms caused
by the above don't trip the limiter while the fix is validated.

Also pin COMPOSE_PROJECT_NAME=testnet in the RPC .env example. Compose
derives the volume prefix from the directory name, so this stack would
look for `rpc_rpc-node-data` while nodes deployed from the old
`docker/testnet/` layout keep their chain DB in `testnet_rpc-node-data`.
Without the pin, migrating an existing RPC starts an archive resync from
block 0.
@nol4lej
nol4lej merged commit 3028a19 into main Jul 28, 2026
1 check passed
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.

1 participant