Ok, that's a good point. I overlooked the possibility of this
considering that I didn't think the timestamp resolution was that high.
I changed my rules and constraints on tables to take care of this case.
Thanks.
However, how does this affect the problem I having considering that
inserts that were getting stuck were not necessarily close to critical
time (midnight)?
Also, today a developer reported similar issue when running a select
statement from this partitioned table. He said that query would just
run forever and if he stopped it and ran it again, it would return data
within seconds. It was behaving in this manner randomly (aka it would
work a few times in a row, then it would get stuck).
As before I am not seeing any locks or unusal behavior in the logs and
pg_stat_activity.