infinite loop on windows graceful restart with proxy websocket #277
Comments
|
I am a bit confused. Does the browser make WebSocket connections over HTTP/2? It should not, if the server does not support it. Is this another request that produces the infinite loop? Strange. Can you produce a log with |
same for me cause my big picture understanding is that in this situation,
no, over HTTP/1.1
just have this relevant: |
|
Hi @icing, To be clear:
But as you said, I’m not sure that the infinite loop is originated by a gracefully restarted websocket cnx!
As it’s difficult to work on Windows with tracing on a “kill & up process fork” and as I’m not fluent with APR, would it be possible to give a tmp patch (better than mine... easy ;) to trace which request are stuck in the infinite loop? |
|
If mod_http2 is indeed polling in a loop, as your stacktrace indicates, something weird is going on. The trace is a call from I think we really need a log from such a case with |
|
Hi @icing,I'm back on it
Changes since last month:
I'll keep you posted... |
|
Thanks for the update. This is a mysterious one indeed. |
|
Hi @icing, I think I've caught the loop
Here is the full 2 minutes httpd-error.log That's the context. If you need more traces, information... just tell me |
|
Hi @icing, I sometime continue diging on this issue and as my dbg is currently “hot” with another bug (php), Here are the new some fresh traces and clues. just to reminde
⏩ Affected thread only hang some h2 (non websocket) connection de be closed (at least I've never seen) Ex. I always have the same call stack : refreshing the stack, line only change in impl_pollset_poll() method This confirm that the condition like https://lists.apache.org/thread/k5sjldo71gw8w9cwov29tol99ybd894z referring to that's where my understanding ends 😊 |
|
Note:
|
|
if the Windows poll always returns EINTR, this would lead to a busy loop. I think you are correct in your read on this. You could try the following patch that will treat EINTR as an error during shutdowns. Index: modules/http2/h2_mplx.c
===================================================================
--- modules/http2/h2_mplx.c (Revision 1921209)
+++ modules/http2/h2_mplx.c (Arbeitskopie)
@@ -1232,9 +1232,15 @@
rv = apr_pollset_poll(m->pollset, timeout >= 0? timeout : -1, &nresults, &results);
H2_MPLX_ENTER_ALWAYS(m);
if (APR_STATUS_IS_EINTR(rv) && m->shutdown) {
+#ifdef WIN32
+ /* See https://github.com/icing/mod_h2/issues/277
+ * WSAPoll seems to always deliver EINTR during server graceful
+ * restarts. Treat this is an error condition */
+#else
if (!m->aborted) {
rv = APR_SUCCESS;
}
+#endif
goto cleanup;
}
} while (APR_STATUS_IS_EINTR(rv)); |
|
Hi @icing, Thx for the feedback!
diff --git a/mod_http2/h2_mplx.c b/mod_http2/h2_mplx.c
index 71a7431..fc9b81e 100644
--- a/mod_http2/h2_mplx.c
+++ b/mod_http2/h2_mplx.c
@@ -1232,10 +1232,17 @@ static apr_status_t mplx_pollset_poll(h2_mplx *m, apr_interval_time_t timeout,
H2_MPLX_LEAVE(m);
rv = apr_pollset_poll(m->pollset, timeout >= 0? timeout : -1, &nresults, &results);
H2_MPLX_ENTER_ALWAYS(m);
+ ap_log_cerror(APLOG_MARK, APLOG_ERR, rv, m->c1, APLOGNO(10310) H2_MPLX_MSG(m, "loop rv=%d eintr=%d shutdown=%d aborted=%d pcount=%d"),(int)rv,APR_STATUS_IS_EINTR(rv),m->shutdown,m->aborted,m->processing_count);
if (APR_STATUS_IS_EINTR(rv) && m->shutdown) {
+#ifdef WIN32
+ /* See https://github.com/icing/mod_h2/issues/277
+ * WSAPoll seems to always deliver EINTR during server graceful
+ * restarts. Treat this is an error condition */
+#else
if (!m->aborted) {
rv = APR_SUCCESS;
}
+#endif
goto cleanup;
}
} while (APR_STATUS_IS_EINTR(rv));
if other information is needed in this line for debugging, just tell me |
Hi @icing
I have a bug with mod_h2 and "proxyfied" websockets when restarting the service (graceful I suppose) on Windows
This problem reminds me of https://bz.apache.org/bugzilla/show_bug.cgi?id=65180 but is not fully identical.
httpd 2.4.58 self compiled vs17 x64 mod_h2 2.0.26
101 Switching Protocolsfor all and I see the exchanges in the console...and to be clear about it...
indeed, log show an infinite loop after graceful restart..
So
H2Upgrade Off, I directly fall into the infinite loopThe text was updated successfully, but these errors were encountered: