Fix: per-pid ust buffers flush race with application unregister

If an application exits between the moment we create the ust app session
(shadow copy) and the moment we send channels to the application,
sending channels may fail. If the application unregisters between the
the end of ust app session creation and the destroy that follows (ust
app session lock is released in between), we may see is_sent being
false.

This assert was initially there in the early days of lttng-tools when
the application was responsible for doing the flush. Now that the flush
has been moved to the consumer daemon, we don't need this assertion
anymore.

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 8c45a89a5f82410e752bf865bbfe889020551ddf..160a15640ff4efd5709af571c9ab3b64b68c0334 100644 (file)
--- a/src/bin/lttng-sessiond/ust-app.c
+++ b/src/bin/lttng-sessiond/ust-app.c
@@ -4469,7 +4469,6 @@ int ust_app_flush_app_session(struct ust_app *app,
                cds_lfht_for_each_entry(ua_sess->channels->ht, &iter.iter, ua_chan,
                                node.node) {
                        health_code_update();
-                       assert(ua_chan->is_sent);
                        ret = consumer_flush_channel(socket, ua_chan->key);
                        if (ret) {
                                ERR("Error flushing consumer channel");