Hi Henning

So I tested. Yesterday I send traffic, around 400 concurrent dialogues over the two DMQ synced instances and stopped the traffic in the evening waiting until today to make sure all calls have ended.

None of the two instances was restarted during that testing.

I have not seen memory related error messages in the logs. So I guess SHM=1024 is large enough now.

This morning:

# kamcmd dlg.stats_active { starting: 0 connecting: 0 answering: 0 ongoing: 0 all: 0 }

So no more running calls on any of the two instances.

But when I look in the local mysql database used to store dialogues:

Instance 2:

mysql> select count(*) from dialog; select count(*) from dialog_vars; +----------+ | count(*) | +----------+ | 0 | +----------+ 1 row in set (0.00 sec)

+----------+ | count(*) | +----------+ | 271 | +----------+ 1 row in set (0.00 sec)

Instance 1:

mysql> select count(*) from dialog; select count(*) from dialog_vars; +----------+ | count(*) | +----------+ | 0 | +----------+ 1 row in set (0.00 sec)

+----------+ | count(*) | +----------+ | 472 | +----------+ 1 row in set (0.00 sec)

So still, quite some variables left in dialog_vars and not the same count. What could be cause this issue?

Mit freundlichen Grüssen

-Benoît Panizzon-

Hello,

thanks for the additional information. It should certainly expire the dialog after some time, even if it's synchronized from DMQ to the first instance.

Maybe you can try it with a small dialog timeout variable in a test setup, if possible. If its not expiring, an issue could be opened about it.

Cheers,

Henning