Set higher default value for Zenoh wait_before_drop parameter

[caguero]

🦟 Bug fix

This patch increases the Zenoh wait_before_drop timeout from 1 ms (Zenoh default) to
50 ms, and max_wait_before_drop_fragments from 50 ms to 250 ms. This
eliminates message loss for large messages in inter-process pub/sub without
measurable throughput impact.

Instructions to reproduce the benchmark

Save this diff into a file:

diff --git a/example/benchmark/plot_benchmark.py b/example/benchmark/plot_benchmark.py
index 5c83ded2..aed7e24e 100755
--- a/example/benchmark/plot_benchmark.py
+++ b/example/benchmark/plot_benchmark.py
@@ -69,10 +69,20 @@ _SIZE_LABELS = [
 # Per-configuration display style.  Order here determines legend order.
 # Wong colorblind-safe palette — distinguishable by hue, even in greyscale.
 STYLES: 'OrderedDict[str, dict]' = OrderedDict([
-    ('zenoh',
-     {'label': 'Zenoh',   'color': '#009E73'}),
+    ('zenoh_default',
+     {'label': 'Zenoh default (1ms)', 'color': '#999999'}),
+    ('zenoh_wait1ms',
+     {'label': 'Zenoh 1ms (explicit)', 'color': '#D55E00'}),
+    ('zenoh_wait5ms',
+     {'label': 'Zenoh 5ms',           'color': '#E69F00'}),
+    ('zenoh_wait10ms',
+     {'label': 'Zenoh 10ms',          'color': '#CC79A7'}),
+    ('zenoh_wait50ms',
+     {'label': 'Zenoh 50ms',          'color': '#009E73'}),
+    ('zenoh_wait100ms',
+     {'label': 'Zenoh 100ms',         'color': '#56B4E9'}),
     ('zeromq',
-     {'label': 'ZeroMQ',  'color': '#0072B2'}),
+     {'label': 'ZeroMQ',              'color': '#0072B2'}),
 ])
 
 
diff --git a/example/benchmark/run_benchmark.py b/example/benchmark/run_benchmark.py
index 647c55ae..84db65d8 100755
--- a/example/benchmark/run_benchmark.py
+++ b/example/benchmark/run_benchmark.py
@@ -73,13 +73,57 @@ import time
 #   name: identifier used in output filenames
 #   env: dict of env vars to SET (keys not present are removed from env)
 #   clear: list of env var names to REMOVE
+_DROP_KEY = ('transport/link/tx/queue/congestion_control/drop/'
+             'wait_before_drop')
+_FRAG_KEY = ('transport/link/tx/queue/congestion_control/drop/'
+             'max_wait_before_drop_fragments')
+
+def _zenoh_wait_config(wait_us, frag_ratio=5):
+    """Build a Zenoh config with a specific wait_before_drop value."""
+    frag_us = wait_us * frag_ratio
+    return {
+        'GZ_TRANSPORT_IMPLEMENTATION': 'zenoh',
+        'GZ_TRANSPORT_ZENOH_CONFIG_OVERRIDE':
+            f'{_DROP_KEY}={wait_us};{_FRAG_KEY}={frag_us}',
+    }
+
 CONFIGS = [
     {
-        'name': 'zenoh',
-        'description': 'Zenoh (default settings)',
+        'name': 'zenoh_default',
+        'description': 'Zenoh (default, wait_before_drop=1ms)',
         'env': {
             'GZ_TRANSPORT_IMPLEMENTATION': 'zenoh',
         },
+        'clear': ['GZ_TRANSPORT_ZENOH_CONFIG_OVERRIDE'],
+    },
+    {
+        'name': 'zenoh_wait1ms',
+        'description': 'Zenoh (wait_before_drop=1ms, explicit)',
+        'env': _zenoh_wait_config(1000),
+        'clear': [],
+    },
+    {
+        'name': 'zenoh_wait5ms',
+        'description': 'Zenoh (wait_before_drop=5ms)',
+        'env': _zenoh_wait_config(5000),
+        'clear': [],
+    },
+    {
+        'name': 'zenoh_wait10ms',
+        'description': 'Zenoh (wait_before_drop=10ms)',
+        'env': _zenoh_wait_config(10000),
+        'clear': [],
+    },
+    {
+        'name': 'zenoh_wait50ms',
+        'description': 'Zenoh (wait_before_drop=50ms)',
+        'env': _zenoh_wait_config(50000),
+        'clear': [],
+    },
+    {
+        'name': 'zenoh_wait100ms',
+        'description': 'Zenoh (wait_before_drop=100ms)',
+        'env': _zenoh_wait_config(100000),
         'clear': [],
     },
     {
@@ -88,7 +132,7 @@ CONFIGS = [
         'env': {
             'GZ_TRANSPORT_IMPLEMENTATION': 'zeromq',
         },
-        'clear': [],
+        'clear': ['GZ_TRANSPORT_ZENOH_CONFIG_OVERRIDE'],
     },
 ]

And then, apply the path and run the benchmark:

cd ~/rotary_ws/src/gz-transport
git apply /tmp/sweep_wait_before_drop.patch
source ~/rotary_ws/install/setup.bash
cd example/benchmark
python3 run_benchmark.py -b build/bench -i 500 -w 10 -t --plot

Summary

Checklist

• Signed all commits for DCO
• Added a screen capture or video to the PR description that demonstrates the fix (as needed)
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• Updated Bazel files (if adding new files). Created an issue otherwise.
• While waiting for a review on your PR, please help review another open pull request to support the maintainers
• Was GenAI used to generate this PR? If so, make sure to add "Generated-by" to your commits. (See this policy for more info.)

Generated-by: Remove this if GenAI was not used.

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by and Generated-by messages.

Backports: If this is a backport, please use Rebase and Merge instead.

[azeey]

Nice! Could you include instructions for the benchmarking you did to discover this so that I or others can reproduce it?

[caguero]

Nice! Could you include instructions for the benchmarking you did to discover this so that I or others can reproduce it?

I've updated the description with the instructions.

[j-rivero]

These can be overridden via ZENOH_CONFIG

Is this correct? the ZENOH configurations seems to be loaded before these lines.

https://github.com/gazebosim/gz-transport/pull/841/changes#diff-652bc237f682bc3650223ea89ba77865c3569765a9943a546f53f655443c9e2fR100

[caguero]

Closing: we'll rely on Zenoh's default congestion-control parameters instead of overriding wait_before_drop. The SHM work continues in #842, now retargeted to main.