diff --git a/2019-04-28-h2load-process-request-failure.html b/2019-04-28-h2load-process-request-failure.html
new file mode 100644
index 0000000..5b2adcf
--- /dev/null
+++ b/2019-04-28-h2load-process-request-failure.html
@@ -0,0 +1,15 @@
+<!--# set var="title" value="h2load Process Request Failure" -->
+<!--# set var="date" value="2019-04-28" -->
+
+<!--# include file="include/top.html" -->
+
+<p>I went looking for an HTTP/2 replacement for <a href="https://httpd.apache.org/docs/2.4/programs/ab.html">Ab</a> and found <a href="https://nghttp2.org/documentation/h2load-howto.html">h2load</a>. Perfect, awesome, just what I need. My first few test runs work fine, then as I'm ramping up test size...</p>
+
+<pre><code>Process Request Failure:1
+</code></pre>
+
+<p>That number at the end seems to be the number of failed requests. Tellingly, it happened after exactly 1000 successful requests each time. Restart h2load, get another 1000 successful requests. I tried <strong>keepalive_requests</strong> in my nginx config, but no luck. Finally, after going through the nginx source for instances of "1000", I found <a href="https://nginx.org/en/docs/http/ngx_http_v2_module.html#http2_max_requests">http2_max_requests</a>, which defaults to 1000. Increase that to an enormous number, restart nginx, everything works fine.</p>
+
+<p>Someone should really implement reconnect in h2load, or at least some decent error messages.</p>
+
+<!--# include file="include/bottom.html" -->
diff --git a/index.html b/index.html
index ace071a..41e6666 100644
--- a/index.html
+++ b/index.html
@@ -3,6 +3,7 @@
 <!--# include file="include/top.html" -->
 
 <ol>
+<li>2019-04-28: <a href="2019-04-28-h2load-process-request-failure.html">h2load Process Request Failure</a></li>
 <li>2019-04-21: <a href="2019-04-21-old-code-roundup.html">Old code roundup</a></li>
 <li>2019-04-14: <a href="2019-04-14-reboot.html">Reboot</a></li>
 <li>2016-05-17: <a href="2016-05-17-wifi-bridging-redux.html">WiFi bridging redux</a></li>
diff --git a/markdown/2019-04-28-h2load-process-request-failure.md b/markdown/2019-04-28-h2load-process-request-failure.md
new file mode 100644
index 0000000..6238189
--- /dev/null
+++ b/markdown/2019-04-28-h2load-process-request-failure.md
@@ -0,0 +1,13 @@
+<!--# set var="title" value="h2load Process Request Failure" -->
+<!--# set var="date" value="2019-04-28" -->
+
+<!--# include file="include/top.html" -->
+
+I went looking for an HTTP/2 replacement for [Ab](https://httpd.apache.org/docs/2.4/programs/ab.html) and found [h2load](https://nghttp2.org/documentation/h2load-howto.html). Perfect, awesome, just what I need. My first few test runs work fine, then as I'm ramping up test size...
+
+	Process Request Failure:1                                                                                                            
+That number at the end seems to be the number of failed requests. Tellingly, it happened after exactly 1000 successful requests each time. Restart h2load, get another 1000 successful requests. I tried **keepalive\_requests** in my nginx config, but no luck. Finally, after going through the nginx source for instances of "1000", I found [http2\_max\_requests](https://nginx.org/en/docs/http/ngx_http_v2_module.html#http2_max_requests), which defaults to 1000. Increase that to an enormous number, restart nginx, everything works fine.
+
+Someone should really implement reconnect in h2load, or at least some decent error messages.
+
+<!--# include file="include/bottom.html" -->
diff --git a/markdown/index.md b/markdown/index.md
index d5aa8c2..b3f376e 100644
--- a/markdown/index.md
+++ b/markdown/index.md
@@ -2,6 +2,7 @@
 
 <!--# include file="include/top.html" -->
 
+1. 2019-04-28: [h2load Process Request Failure](2019-04-28-h2load-process-request-failure.html)
 1. 2019-04-21: [Old code roundup](2019-04-21-old-code-roundup.html)
 1. 2019-04-14: [Reboot](2019-04-14-reboot.html)
 1. 2016-05-17: [WiFi bridging redux](2016-05-17-wifi-bridging-redux.html)