Alexander Kukushkin

@jberkus It is not hard to implement such hook. If we get non zero exit code from the script, we will just dynamically set `nofailover` tag to `True`. But I...

You have to examine Patroni and Postgres logs in order to understand what is going on.

You also have to double-check that the replica is streaming from the leader. Replication lag=0 doesn't guarantee that.

Well, auto-pilot will not help if you don't know how to fly an airplane manually. Here is the same story, you need to learn about streaming replication, clustering and HA...

Fairly speaking I have never ever seen a situation when the postmaster hung so that is not able to respond at all, but I have seen a lot of situations...

> Unfortunately, we ran into this issue on both databases Oracle & PostgreSQL. I can't tell anything about Oracle... We are running hundreds of postgres clusters and statistically we would...

> I gave a scenario where only the postmaster was hung Sorry, but no. When you send `SIGSTOP` you just shoot your own foot, and it hurts badly. > Looks...

> But why the subprocess.call doesn't return? Is it related to the async call? I don't see any single reason for such behavior. > Patroni version: 2.0.2 No matter what...

Something is fishy. The `Reaped pid=` messages you are supposed to see only if Patroni is started in the container with pid=1. This is very unusual and should not be...

> I think Patroni should support that, no? As I already said, we do our best, but no guarantees. Please use dump-init.