Project

General

Profile

Bug #1206

live per-pid channel with short lived app becomes unusable

Added by Francis Deslauriers about 2 months ago.

Status:
New
Priority:
Normal
Assignee:
-
Target version:
-
Start date:
10/29/2019
Due date:
% Done:

0%

Estimated time:

Description

When launching a short lived UST app in a per-pid session, the sessiond gets an error on the consumer socket and seems to forget the channel configuration. Note that in this scenario, I did not connect to the relayd with Babeltrace but if I try afterwards BT don't see any session to connect to.

Here is my channel configuration and how I launched the app:

lttng create claudia_normand_2 --live
lttng enable-channel --buffers-pid -u chan1
lttng enable-event -u -a -c chan1
lttng start
./a.out &

After running this I can't see the channel configuration when I run lttng status:

Tracing session claudia_normand_2: [active]
    Trace output: tcp4://127.0.0.1:5342/ [data: 5343]

=== Domain: UST global ===

Buffer type: per PID

Here are the errors in the sessiond log:

DEBUG1 - 14:47:53.804518546 [13785/13822]: End of waiter wait period (in lttng_waiter_wait() at waiter.c:101)
DEBUG3 - 14:47:53.804666572 [13785/13822]: Session 0 found by id. (in session_find_by_id() at session.c:952)
DEBUG3 - 14:47:53.804681826 [13785/13822]: Consumer discarded events id 0 (in consumer_get_discarded_events() at consumer.c:1577)
DEBUG1 - 14:47:53.939490348 [13785/13813]: Error when receiving data from the consumer socket 62 (in consumer_socket_recv() at consumer.c:166)
Error: Handling metadata request
PERROR - 14:47:53.939490345 [13785/13822]: recvmsg: Connection reset by peer (in lttcomm_recv_unix_sock() at unix.c:210)
DEBUG1 - 14:47:53.939556026 [13785/13822]: Error when receiving data from the consumer socket 61 (in consumer_socket_recv() at consumer.c:166)
Error: get discarded events
DEBUG2 - 14:47:53.939577920 [13785/13822]: Trace UST channel chan1 found by name (in trace_ust_find_channel_by_name() at trace-ust.c:192)
DEBUG3 - 14:47:53.939616589 [13785/13822]: Buffer registry per PID find id: 1 (in buffer_reg_pid_find() at buffer-registry.c:318)
DEBUG3 - 14:47:53.939624603 [13785/13822]: Buffer registry per PID destroy with id: 1 (in buffer_reg_pid_destroy() at buffer-registry.c:740)
DEBUG2 - 14:47:53.939648741 [13785/13822]: UST app pid 13827 deleted (in delete_ust_app() at ust-app.c:940)
DEBUG3 - 14:47:53.939661666 [13785/13787]: [ht-thread] Returning from poll on 2 fds. (in thread_ht_cleanup() at ht-cleanup.c:136)
Error: Health error occurred in thread_consumer_management
DEBUG1 - 14:47:53.939749103 [13785/13813]: consumer thread cleanup completed (in thread_consumer_management() at manage-consumer.c:420)
DEBUG1 - 14:47:53.939757882 [13785/13813]: Thread "Consumer management" has returned (in launch_thread() at thread.c:76)

I am using this version of the lttng: lttng (LTTng Trace Control) 2.12.0-pre - Codename TBD - v2.11.0-rc1-549-g93d0d1f7d

I attached the full log of the sessiond and relayd.


Files

relayd.log (164 KB) relayd.log Francis Deslauriers, 10/29/2019 02:55 PM
sessiond.log (200 KB) sessiond.log Francis Deslauriers, 10/29/2019 02:55 PM

Also available in: Atom PDF