Account the lost packets in snapshot mode

When working in flight-recorder, we cannot rely on the absolute sequence
number because we only extract the content of the ring buffer
periodically. Moreover, we don't want to account the lost packets at the
beginning or the end of the snapshot since the "holes" don't appear in
the resulting snapshot, so we make sure we only account lost packets
located between extracted packets.

Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

diff --git a/src/common/kernel-consumer/kernel-consumer.c b/src/common/kernel-consumer/kernel-consumer.c
index 2ea5fa114223f101362e4d780d106b25b3812238..42d471d1a9a040b10e6ee48b56d66f34544f8091 100644 (file)
--- a/src/common/kernel-consumer/kernel-consumer.c
+++ b/src/common/kernel-consumer/kernel-consumer.c
@@ -141,6 +141,8 @@ int lttng_kconsumer_snapshot_channel(uint64_t key, char *path,
        }
 
        cds_list_for_each_entry(stream, &channel->streams.head, send_node) {
+               /* Are we at a position _before_ the first available packet ? */
+               bool before_first_packet = true;
 
                health_code_update();
 
@@ -227,6 +229,7 @@ int lttng_kconsumer_snapshot_channel(uint64_t key, char *path,
                while (consumed_pos < produced_pos) {
                        ssize_t read_len;
                        unsigned long len, padded_len;
+                       int lost_packet = 0;
 
                        health_code_update();
 
@@ -241,6 +244,15 @@ int lttng_kconsumer_snapshot_channel(uint64_t key, char *path,
                                }
                                DBG("Kernel consumer get subbuf failed. Skipping it.");
                                consumed_pos += stream->max_sb_size;
+
+                               /*
+                                * Start accounting lost packets only when we
+                                * already have extracted packets (to match the
+                                * content of the final snapshot).
+                                */
+                               if (!before_first_packet) {
+                                       lost_packet = 1;
+                               }
                                continue;
                        }
 
@@ -284,6 +296,16 @@ int lttng_kconsumer_snapshot_channel(uint64_t key, char *path,
                                goto end_unlock;
                        }
                        consumed_pos += stream->max_sb_size;
+
+                       /*
+                        * Only account lost packets located between
+                        * succesfully extracted packets (do not account before
+                        * and after since they are not visible in the
+                        * resulting snapshot).
+                        */
+                       stream->chan->lost_packets += lost_packet;
+                       lost_packet = 0;
+                       before_first_packet = false;
                }
 
                if (relayd_id == (uint64_t) -1ULL) {

diff --git a/src/common/ust-consumer/ust-consumer.c b/src/common/ust-consumer/ust-consumer.c
index d0f7ac53ee651371ba4b59444df5a8ae6c391330..5b4b1496556ad298349e407a7e077413d758a121 100644 (file)
--- a/src/common/ust-consumer/ust-consumer.c
+++ b/src/common/ust-consumer/ust-consumer.c
@@ -1029,6 +1029,8 @@ static int snapshot_channel(uint64_t key, char *path, uint64_t relayd_id,
        DBG("UST consumer snapshot channel %" PRIu64, key);
 
        cds_list_for_each_entry(stream, &channel->streams.head, send_node) {
+               /* Are we at a position _before_ the first available packet ? */
+               bool before_first_packet = true;
 
                health_code_update();
 
@@ -1095,6 +1097,7 @@ static int snapshot_channel(uint64_t key, char *path, uint64_t relayd_id,
                while (consumed_pos < produced_pos) {
                        ssize_t read_len;
                        unsigned long len, padded_len;
+                       int lost_packet = 0;
 
                        health_code_update();
 
@@ -1108,6 +1111,15 @@ static int snapshot_channel(uint64_t key, char *path, uint64_t relayd_id,
                                }
                                DBG("UST consumer get subbuf failed. Skipping it.");
                                consumed_pos += stream->max_sb_size;
+
+                               /*
+                                * Start accounting lost packets only when we
+                                * already have extracted packets (to match the
+                                * content of the final snapshot).
+                                */
+                               if (!before_first_packet) {
+                                       lost_packet = 1;
+                               }
                                continue;
                        }
 
@@ -1143,6 +1155,16 @@ static int snapshot_channel(uint64_t key, char *path, uint64_t relayd_id,
                                goto error_close_stream;
                        }
                        consumed_pos += stream->max_sb_size;
+
+                       /*
+                        * Only account lost packets located between
+                        * succesfully extracted packets (do not account before
+                        * and after since they are not visible in the
+                        * resulting snapshot).
+                        */
+                       stream->chan->lost_packets += lost_packet;
+                       lost_packet = 0;
+                       before_first_packet = false;
                }
 
                /* Simply close the stream so we can use it on the next snapshot. */