You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@pulsar.apache.org by Apache Pulsar Slack <ap...@gmail.com> on 2018/08/25 09:11:02 UTC
Slack digest for #general - 2018-08-25
2018-08-24 09:41:32 UTC - Matti-Pekka Laaksonen: Thank you for the help yesterday, I was now able to get zookeeper and broker/bookie nodes up and somewhat running
----
2018-08-24 09:42:22 UTC - Matti-Pekka Laaksonen: I have not tried to connect to the brokers with a pulsar client yet, but currently the bookies and zookeepers behave very inconsistently
----
2018-08-24 09:43:35 UTC - Matti-Pekka Laaksonen: I have reinstalled the stack a few times, and sometimes I'm able to initialize the cluster metadata on a zookeeper node just fine, and some times I get an error stating that the node already exists, which suggests that the metadata has been initialized already
----
2018-08-24 09:45:50 UTC - Matti-Pekka Laaksonen: And when trying the "bookiesanity" and bookie "simpletest" tests, they some times work just fine and sometimes throw a runtime exception indicating that other bookies can not be found, which I would guess is due to problems connecting to the zk cluster
----
2018-08-24 09:46:20 UTC - Matti-Pekka Laaksonen:
----
2018-08-24 09:48:21 UTC - Matti-Pekka Laaksonen: When I try to get the zk stats from a broker node ("echo stat | nc <zk node ip> 2181") I mostly get a proper answer, but sometimes no answer at all
----
2018-08-24 09:48:32 UTC - Ali Ahmed: looks to be your node addresses are not stable and are changing frequently
----
2018-08-24 09:50:10 UTC - Matti-Pekka Laaksonen: But they are not. All the nodes belong to the same internal network, and the IP addresses don't change
----
2018-08-24 09:53:56 UTC - Matti-Pekka Laaksonen: This inconsistency is what really bugs me, and I don't really know where to start looking
----
2018-08-24 11:18:27 UTC - Matti-Pekka Laaksonen: Actually, I think the problem is with the tests and Bookkeeper. The pulsar cluster itself seems to be (almost) working
----
2018-08-24 11:18:57 UTC - Matti-Pekka Laaksonen: I can use the admin tool to list brokers, add a tenant and add a namespace
----
2018-08-24 11:19:43 UTC - Matti-Pekka Laaksonen: But when I try to produce a single message with the pulsar-client (sudo bin/pulsar-client produce <persistent://public/default/test> -n 1 -m "Hello, Pulsar"
) I get the following error message:
----
2018-08-24 11:20:56 UTC - Matti-Pekka Laaksonen:
----
2018-08-24 11:29:41 UTC - Matti-Pekka Laaksonen: Are there other steps besides the "initialize cluster metadata" I have to do before I can produce messages to a topic?
----
2018-08-24 12:04:39 UTC - Matti-Pekka Laaksonen: I tried to update the tenant with the pulsar-admin tool, adding my cluster as the allowed cluster and "all" as admin roles. Still, I get the same error message
----