Slack digest for #general - 2019-07-21

[Apache Pulsar Slack]

2019-07-20 09:56:11 UTC - Gurgen Hovhannisyan: Hi guys.

I am having issues by deleting schemas (you can see topic is deleted):
./pulsar-admin persistent delete <persistent://public/default/rssi_data_topic>
...........................
    Reason: Topic not found
./pulsar-admin schemas delete  <persistent://public/default/rssi_data_topic>
./pulsar-admin schemas get rssi_data_topic
{
  "name": "rssi_data_topic",
................

you can see after deleting topic and schema with that name, I still see the 
schema, why ? how to completely delete all my schemas &amp; topics so I don't 
see any of them.

this causes consumer to not work at all, as it always shows error: incompatible 
schema
----
2019-07-20 10:39:47 UTC - Nicolas Ha: How can I start pulsar SQL and configure 
it via environment variables? In docker-compose for instance I tried this:
```
    environment:
      - PULSAR_PREFIX_BROKER_SERVICE_URL=<http://pulsarstandalone:8080>
    command: &gt;
      /bin/bash -c
      "bin/apply-config-from-env.py conf/standalone.conf
      &amp;&amp; bin/pulsar sql-worker run"
```
but it doesn’t has any affect and when starting it I see
```
2019-07-20T10:38:06.978Z        INFO    main    Bootstrap       PROPERTY        
                           DEFAULT                                        
RUNTIME                                        DESCRIPTION
2019-07-20T10:38:06.978Z        INFO    main    Bootstrap       
pulsar.broker-service-url                  <http://localhost:8080>              
            <http://localhost:8080>
```
(I tried using conf/presto/catalog/pulsar.properties too )
----
2019-07-20 15:01:11 UTC - Matteo Merli: @Nicolas Ha
&gt;       “bin/apply-config-from-env.py conf/standalone.conf

The problem this with python script is that it looks for env variables matching 
the config key. In the presto config `pulsar.properties`, the key names are 
like:

```
pulsar.broker-service-url=<http://localhost:8080>
...
```

I think the `.` and `-` are the problem here, since they’re not valid as env 
variable names.

The easiest solution would be to mount the config file directly into the 
container.
----
2019-07-20 15:25:40 UTC - Nicolas Ha: Oh I see - that would work in dev, but 
how do people deploy it on kubernetes then?
I usually use env vars there. So you would have to build an image with the 
file, or is there another way?
----
2019-07-20 15:26:43 UTC - Matteo Merli: In K8S you can pass the file as a 
ConfigMap, which can be mounted as a file in the pod
+1 : Nicolas Ha, Ali Ahmed
----