2020-02-02 12:51:41 UTC - Alex Yaroslavsky: This 
<https://github.com/apache/pulsar/issues/6131> helped (setting BOOKIE_MEM)
----
2020-02-02 13:50:18 UTC - rockybean: @rockybean has joined the channel
----
2020-02-02 15:37:10 UTC - Andrew Tan: @Andrew Tan has joined the channel
----
2020-02-02 16:36:27 UTC - David Kjerrumgaard: @ravi satya durga prasad Yenugula 
It is the last line of the docker-compose.yml file. Sorry for the confusion.
----
2020-02-02 16:41:28 UTC - Guilherme Perinazzo: So why do clients keep track of 
the which messages have been acked before sending it to the broker? Shouldn't 
it just ack based on the batch index?
----
2020-02-02 16:58:05 UTC - Christoffer Soop: @Christoffer Soop has joined the 
channel
----
2020-02-02 17:00:53 UTC - Guilherme Perinazzo: The pure go client doesn't send 
the batch index when acking to the broker
----
2020-02-02 17:21:18 UTC - Sijie Guo: The limitation is at the broker side right 
now. Because broker tracks cursors at entry level. There is an effort on making 
broker track cursors at message level. 
----
2020-02-02 17:22:49 UTC - Sijie Guo: Yes you need to set BOOKIE_MEM
----
2020-02-02 17:23:07 UTC - Guilherme Perinazzo: So there's currently no way to 
ack an individual message inside a batch? I'm asking because I'm trying to 
implement batch message handling on the rust client
----
2020-02-03 00:42:21 UTC - Greg Methvin: As I understand it, currently an ack 
within a batch is recorded purely on the client side. Once all messages in the 
batch are acked by the client, the entire batch is acknowledged. If any of of 
the batch messages are nacked or remain unacked after the ack timeout, the 
entire batch is redelivered.
----
2020-02-03 00:44:17 UTC - Greg Methvin: This is one of the related bugs: 
<https://github.com/apache/pulsar/issues/5969>
----
2020-02-03 03:40:32 UTC - Andrew Tan: Hi there, I have a question about the 
Client APIs specifically Python and Websocket. I'm looking to perform an 
operation where I pull all message since a certain timestamp (assume the 
timestamp is within the retention period) is this possible in the API directly?
----
2020-02-03 04:17:40 UTC - sky-big: @sky-big has joined the channel
----
2020-02-03 04:21:43 UTC - Eugen: @Andrew Tan The reader interface 
<https://pulsar.apache.org/docs/en/concepts-clients/#reader-interface> allows 
you to specify a message id from where to start. If all you have is a 
timestamp, you will first have to figure out the message id e.g. using Pulsar 
SQL,:
```SELECT * FROM pulsar."public/default"."my-topic" WHERE __publish_time__ &gt; 
timestamp '2020-01-22' LIMIT 1;```
which will result in something like this:
```      __value__       | __partition__ | __event_time__ |    __publish_time__ 
    | __message_id__ | __sequence_id__ | __producer_name__ | __key__ |    
__properties__    
----------------------+---------------+----------------+-------------------------+----------------+-----------------+-------------------+---------+----------------------
 6d 79 76 61 6c 75 65 |            -1 | NULL           | 2020-01-22 
14:51:27.227 | (612,0,0)      |               0 | standalone-0-2    | mykey   | 
{"prop1":"propVal1"} 
(1 row)```
----
2020-02-03 06:38:50 UTC - Manju Priya A R: @Manju Priya A R has joined the 
channel
----
2020-02-03 07:18:03 UTC - Manju Priya A R: Hi, Is there any specific reason on 
why max-retries and dead letter topic is not supported for sinks while it is 
available in functions?
----
2020-02-03 08:24:02 UTC - Sijie Guo: no. it is probably not added to sinks yet.
----

Reply via email to