[lttng-dev] Lttng 2.6 relayd hanged
Jonathan Rajotte Julien
Jonathan.rajotte-julien at efficios.com
Mon Jul 6 11:15:07 EDT 2015
In addition to what Jérémie asked:
On 2015-07-03 05:32 PM, kranthi kumar wrote:
> Dear Sir/Madam,
> We are using Lttng 2.6
Are you building from source, git? Could you run "lttng --version" on
all systems and paste the result.
> We are using Lttng for tracing across multiple systems.
> We have a relayd running on a remote-host with several system(4) with
> 2 active sessions each, generating traces towards this remote-host.
> After running for 2-3 days, we observed that the tracing systems
> restarted and the sessions are killed.
Is there any logs of this? If not could you retry the experiment with a
lttng-sessiond with the "-vvv" (verbose mode) flag set and log the
ouptut to a file ( 2>&1 in (ba)sh ). Make sure that the newly spawned
relayd/sessiond does not overwrite the previous log.
This option can be used on the relayd also to produce useful information.
> However we observed sockets of relayd in ESTABLISHED state. We also
> verified that ps command showing relayd active. From “netstat –tanp”
> we confirmed that relayd is listening on default port.
> When we tried to establish new sessions and start streaming towards
> this relayd from our trace generating systems(all 4 one at a time)
> again, all the systems hanged indefinitely without showing any error
Here the "-vvv" flag on any "lttng" command might show some information.
If you are able to reproduce could you also provide us the verbose
output of "lttng create" command and the corresponding log from
lttng-sessiond and lttng-relayd if any.
> The only way to recover our trace generating systems is to restart the
> relayd on remote host. If you restart the relayd and then create new
> sessions, streaming can be started again.
> Please let us know if this is a known issue.
> lttng-dev mailing list
> lttng-dev at lists.lttng.org
Jonathan R. Julien
More information about the lttng-dev