Scripting and automation

Every field in the wizard is a flag, so anything you work out interactively can be pasted into a script.

The three ways to start

Command	Wizard	Dashboard	Use for
`wireblast`	yes	yes	Exploring, first runs
`wireblast --start ...`	no	yes	You know the settings, you want to watch
`wireblast --no-tui ...`	no	no	Scripts, CI, `nohup`, no terminal at all

--no-tui validates the flags, runs, and prints plain text to stdout. No terminal required.

sudo wireblast --no-tui -i eno2 --dst-ip 192.0.2.10 \
  --packet-size 512 --pps 100k -d 10s -y

What it prints

A plan, a start line, one line per second, and a summary:

destination   192.168.0.2/32 port 9000
next hop      3c:ec:ef:b4:c2:dc (given with --dst-mac)
flows         64
rate          200kpps (aggregate across queues)
duration      12s

attaching XDP to eno2 and waiting for the link...
link came back after 11.8s
started: eno2: 12 queue(s), zero-copy, native XDP, driver ixgbe, rx filter none
running for 12s (Ctrl-C to stop early)

[0:01] tx 149.89 k pkts  199.71 kpps  L1 610.02 Mbit/s  L2 578.07 Mbit/s  avg 362B
[0:02] tx 350.02 k pkts  200 kpps  L1 610.95 Mbit/s  L2 578.95 Mbit/s  avg 362B

ran for 0:12
  tx: 2.4 M packets, 868.24 MB, 199.86 kpps, L1 610.52 Mbit/s, L2 578.54 Mbit/s, avg frame 362B
      udp 2.4 M, tcp 0, other 0

There’s no --json yet. Parsing is text parsing for now, and the ran for summary block is the stable thing to key on.

Extracting the totals:

sudo wireblast --no-tui -i eno2 --dst-ip 192.0.2.10 --pps 1M -d 30s -y \
  | awk '/^  tx:/ {print $2, $3}'

Exit codes

Code	Meaning
`0`	Ran to completion, or was stopped cleanly with Ctrl-C
`1`	Something went wrong. The reason is on stderr, prefixed `wireblast:`
`130`	A second interrupt while shutting down, the emergency exit

Errors go to stderr, statistics to stdout, so you can separate them:

sudo wireblast --no-tui ... 2>/var/log/wireblast.err | tee /var/log/wireblast.out

Confirmations

Some runs stop and ask. There are two levels, and they behave differently on purpose. Ordinary confirmations cover things like your SSH session being on this interface, or unlimited rate on the default-route interface. --yes (or -y) answers these:

sudo wireblast --no-tui -i eno2 --pps unlimited -d 30s --yes

The match-all guard covers --rx-mode all. --yes does not answer this one. You need --allow-match-all as well:

sudo wireblast --no-tui -i eth1 --mode receive \
  --rx-mode all --allow-match-all --yes -d 60s

This is deliberate. --yes means “I’ve read the warnings”. --allow-match-all means “I specifically accept that every packet on this interface stops reaching the kernel, including SSH”. They’re different statements and Wireblast makes you make both.

Without a terminal and without the flags, it refuses rather than hanging:

this run needs an explicit confirmation but there is no terminal to ask on.
Re-run it interactively, or pass --yes together with --allow-match-all if you
have read the warnings above

Saved settings don’t apply

--no-tui neither reads nor writes the saved settings in ~/.wireblast/. A scripted run depends only on its flags, so it behaves identically on a fresh machine and on one you’ve been experimenting on all afternoon. If you’re scripting the TUI form (--start) and want to ignore whatever was saved:

sudo wireblast --forget

That clears the files and starts from defaults.

Running unattended

sudo nohup wireblast --no-tui -i eno2 --dst-ip 192.0.2.10 \
  --pps 1M -d 3600s -y > /var/log/wireblast.log 2>&1 &

Wireblast handles SIGINT and SIGTERM cleanly: it drains the transmit rings, collects final counts, detaches the XDP program, and exits 0. So kill on that PID gives you a proper summary rather than a truncated log. A second signal exits immediately with 130, so a wedged run can still be killed without reaching for another shell.

--no-tui attaches and detaches per run, so each invocation pays the link bounce. Looping many short runs in a shell script costs about 8 seconds each on a physical NIC. If you’re iterating, one longer run, or the TUI’s r key, is much faster.

A two-box script

#!/usr/bin/env bash
set -euo pipefail

RECEIVER=192.0.2.20
IFACE=eno2
DUR=30

ssh root@$RECEIVER "nohup wireblast --no-tui -i $IFACE --mode receive \
  --rx-mode udp-port --rx-port 9000 -d $((DUR + 20))s -y \
  > /tmp/rx.log 2>&1 &"

sleep 2

sudo wireblast --no-tui -i $IFACE --dst-ip 192.0.2.20 --dst-port 9000 \
  --packet-size 512 --pps 1M -d ${DUR}s -y

ssh root@$RECEIVER "until grep -q '^ran for' /tmp/rx.log; do sleep 2; done; \
  grep -A3 '^ran for' /tmp/rx.log"

Give the receiver a longer duration than the sender so it’s listening for the whole run, and start it first. Packets sent before it attaches are simply gone. The examples directory has ready-made scripts for most of this, all driven by environment variables.

Getting started

Concepts

Traffic patterns

Guides

Reference

Scripting and automation

The three ways to start

What it prints

Exit codes

Confirmations

Saved settings don’t apply

Running unattended

A two-box script

​The three ways to start

​What it prints

​Exit codes

​Confirmations

​Saved settings don’t apply

​Running unattended

​A two-box script

The three ways to start

What it prints

Exit codes

Confirmations

Saved settings don’t apply

Running unattended

A two-box script