Fix: metadata push -EPIPE should be recoverable

This return value can be caused by application terminating concurrently
(when using per-PID buffers), so it should not make the consumer
management thread exit.

CC: Aravind HT <aravind.ht@gmail.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

diff --git a/src/bin/lttng-sessiond/ust-app.c b/src/bin/lttng-sessiond/ust-app.c
index 8508a6e7c286f954e475422638e239ecbc833c58..5b9c514976cc640e9db7f41e1251a26de9a81485 100644 (file)
--- a/src/bin/lttng-sessiond/ust-app.c
+++ b/src/bin/lttng-sessiond/ust-app.c
@@ -430,6 +430,9 @@ void delete_ust_app_channel(int sock, struct ust_app_channel *ua_chan,
  * Must be called with the registry lock held.
  *
  * On success, return the len of metadata pushed or else a negative value.
+ * Returning a -EPIPE return value means we could not send the metadata,
+ * but it can be caused by recoverable errors (e.g. the application has
+ * terminated concurrently).
  */
 ssize_t ust_app_push_metadata(struct ust_registry_session *registry,
                struct consumer_socket *socket, int send_zero_data)
@@ -453,9 +456,10 @@ ssize_t ust_app_push_metadata(struct ust_registry_session *registry,
        /*
         * On a push metadata error either the consumer is dead or the
         * metadata channel has been destroyed because its endpoint
-        * might have died (e.g: relayd). If so, the metadata closed
-        * flag is set to 1 so we deny pushing metadata again which is
-        * not valid anymore on the consumer side.
+        * might have died (e.g: relayd), or because the application has
+        * exited. If so, the metadata closed flag is set to 1 so we
+        * deny pushing metadata again which is not valid anymore on the
+        * consumer side.
         */
        if (registry->metadata_closed) {
                return -EPIPE;
@@ -546,6 +550,9 @@ error_push:
  * of socket throughout this function.
  *
  * Return 0 on success else a negative error.
+ * Returning a -EPIPE return value means we could not send the metadata,
+ * but it can be caused by recoverable errors (e.g. the application has
+ * terminated concurrently).
  */
 static int push_metadata(struct ust_registry_session *registry,
                struct consumer_output *consumer)

diff --git a/src/bin/lttng-sessiond/ust-consumer.c b/src/bin/lttng-sessiond/ust-consumer.c
index f77f67d7f4731333bd44088554e0768fce85685a..75d7170d2752f4d952d8d8a69bd27eed0243d565 100644 (file)
--- a/src/bin/lttng-sessiond/ust-consumer.c
+++ b/src/bin/lttng-sessiond/ust-consumer.c
@@ -490,12 +490,15 @@ int ust_consumer_metadata_request(struct consumer_socket *socket)
        pthread_mutex_lock(&ust_reg->lock);
        ret_push = ust_app_push_metadata(ust_reg, socket, 1);
        pthread_mutex_unlock(&ust_reg->lock);
-       if (ret_push < 0) {
+       if (ret_push == -EPIPE) {
+               DBG("Application or relay closed while pushing metadata");
+       } else if (ret_push < 0) {
                ERR("Pushing metadata");
                ret = -1;
                goto end;
+       } else {
+               DBG("UST Consumer metadata pushed successfully");
        }
-       DBG("UST Consumer metadata pushed successfully");
        ret = 0;
 
 end: