Skip to content

demo: document kernel rcvbuf sysctls for 64 MiB parser buffer #23

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

Open
armcconnell wants to merge 1 commit into
base: main
Choose a base branch
from
Open

demo: document kernel rcvbuf sysctls for 64 MiB parser buffer #23

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
18 changes: 18 additions & 0 deletions demo/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,24 @@ Everything after the feed runs in Docker. The parser uses host networking (for I

The demo is designed for a host where the DoubleZero tunnel is already in place.

## Host tuning (kernel socket buffers)

The parsers raise their UDP receive buffer to 64 MiB to absorb feed bursts, but
the kernel silently clamps that request to `net.core.rmem_max`. On a fresh host
the default ceiling is far lower, so the 64 MiB is only partially honored and
sustained bursts can still drop packets at the socket layer. Before running
under a live feed, raise these sysctls (persist them in `/etc/sysctl.d/`):

```bash
sysctl -w net.core.rmem_max=134217728
sysctl -w net.core.rmem_default=16777216
sysctl -w net.core.netdev_max_backlog=50000
```

The parser decouples its unix sink from the UDP read path so a slow consumer
can't back-pressure the socket, but that headroom only helps if the kernel
actually grants the larger buffer — hence the sysctls above.

## Quick start

```bash
Expand Down
Loading