OK, looking closer at the suggested patch I see it fixes
trace2-Loggging. When I activate e.g. trace8, I do get the expected
crash in ap_proxy_check_connection.
@Konstantin: do you have LogLevel trace2 or higher? If s, do you also
see crashes with trace1 or lower?
@Yann: your patch is just to check whether that is the problem? I guess
the final patch will distinguish between conn->connection being NULL
(then using your suggested code) or non-NULL (then using the original
code). Or would you always use your suggested code?
Thanks and regards,
Rainer
Am 22.12.2016 um 18:38 schrieb Yann Ylavic:
On Thu, Dec 22, 2016 at 3:12 PM, Jim Jagielski <j...@jagunet.com> wrote:
Hmmmm...
I wonder if it's due to:
/* Step Two: Make the Connection */
- if (ap_proxy_connect_backend(scheme, backend, worker, r->server)) {
+ if (ap_proxy_check_connection(scheme, backend, r->server, 0,
+ PROXY_CHECK_CONN_EMPTY)
+ && ap_proxy_connect_backend(scheme, backend, worker,
+ r->server)) {
ap_log_rerror(APLOG_MARK, APLOG_ERR, 0, r, APLOGNO(00896)
Probably, I think it comes from :
Index: modules/proxy/proxy_util.c
===================================================================
--- modules/proxy/proxy_util.c (revision 1775191)
+++ modules/proxy/proxy_util.c (working copy)
@@ -2754,10 +2754,11 @@ PROXY_DECLARE(apr_status_t) ap_proxy_check_connect
}
if (rv == APR_SUCCESS) {
+ apr_sockaddr_t *local_addr = NULL;
+ apr_socket_addr_get(&local_addr, APR_LOCAL, conn->sock);
ap_log_error(APLOG_MARK, APLOG_TRACE2, 0, server,
"%s: reusing backend connection %pI<>%pI",
- scheme, conn->connection->local_addr,
- conn->connection->client_addr);
+ scheme, local_addr, conn->addr);
}
else if (conn->sock) {
/* This clears conn->scpool (and associated data), so backup and
_
because AJP has no conn->connection (conn->sock only).
Konstantin, can you apply a patch?
Regards,
Yann.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@httpd.apache.org
For additional commands, e-mail: users-h...@httpd.apache.org