128GB of ram should be enough to process 1M events. Could you show systemctl status yeti-cdr-billing@cdr_billing
and last lines of dmesg
If you want drop events you should connect to CDR database and run:
cdr=# select * from pgq.get_consumer_info();
queue_name | consumer_name | lag | last_seen | last_tick | current_batch | next_tick | pending_events
-------------+---------------+----------------+-----------------+-----------+---------------+-----------+----------------
cdr_billing | cdr_billing | 09:48:56.63222 | 00:00:02.232517 | 1199607 | | | 0
(1 row)
cdr=# select pgq.unregister_consumer('cdr_billing','cdr_billing');
unregister_consumer
---------------------
1
(1 row)
cdr=# select * from pgq.get_consumer_info();
queue_name | consumer_name | lag | last_seen | last_tick | current_batch | next_tick | pending_events
------------+---------------+-----+-----------+-----------+---------------+-----------+----------------
(0 rows)
cdr=# select pgq.register_consumer('cdr_billing','cdr_billing');
register_consumer
-------------------
1
(1 row)
cdr=# select * from pgq.get_consumer_info();
queue_name | consumer_name | lag | last_seen | last_tick | current_batch | next_tick | pending_events
-------------+---------------+-----------------+----------------+-----------+---------------+-----------+----------------
cdr_billing | cdr_billing | 09:49:21.552044 | 00:00:08.12832 | 1199607 | | | 0
(1 row)
cdr=#
then you should restart cdr_billing process.