2018-10-25 11:39:35 UTC - Alexander Tceluiko: @Alexander Tceluiko has joined
the channel
----
2018-10-25 12:55:00 UTC - David Tinker: @David Tinker has joined the channel
----
2018-10-25 13:03:36 UTC - George Wilk: Trying to set TTL using REST API: POST
/admin/v2/namespaces/:tenant/:namespace/messageTTL
What should be the payload of the POST request? I tried unsuccessfully various
combinations of
`curl -X POST -H 'Content-Type: application/json' --data '{"messageTTL":"120"}'
<https://pulsar-broker:8443/admin/v2/namespaces/mytenant/mynamespace/messageTTL>`
----
2018-10-25 13:08:24 UTC - David Tinker: Anyone running ./bin/pulsar standalone
on OS X Mojave? I upgraded and now Pulsar isn't working. I was using
2.1.1-incubating and have also tried 2.2.0.
----
2018-10-25 13:10:24 UTC - David Tinker: There are no obvious errors on the
console but it doesn't end up listening on its ports ($ netstat -an -ptcp |
grep LISTEN)
----
2018-10-25 13:17:21 UTC - Sijie Guo: the payload should just one integer. so
try `--data '120'`
----
2018-10-25 13:24:38 UTC - David Tinker: Sorted! I was using an old Java 8
(jdk1.8.0_31.jdk) and upgrading to jdk1.8.0_192.jdk got things working.
----
2018-10-25 13:25:17 UTC - Sijie Guo: interesting, jdk version shouldn’t be a
problem though
----
2018-10-25 13:49:36 UTC - Sanjeev Kulkarni: @David Tinker thats interesting. I
upgraded to Mojava a couple of weeks ago and didnt notice this issue.
----
2018-10-25 14:33:05 UTC - Davor Poldrugo: @Davor Poldrugo has joined the channel
----
2018-10-25 14:39:26 UTC - George Wilk: Thanks! This worked well! Is there a
way to specify time units if you want to set TTL to say - 2 weeks?
----
2018-10-25 15:02:06 UTC - Rodrigo Malacarne: Hello, reading the docs about
topic compaction it states that “currently only Java clients can consume
messages from compacted topics”... as there is no specific date or Pulsar
version in this doc page, my question is: is this still valid? Or is it already
possible to consume from compacted topics using Python or C++ consumers?
----
2018-10-25 15:02:54 UTC - Rodrigo Malacarne: If not, is this being planned for
Pulsar v2.2?
----
2018-10-25 15:06:22 UTC - Sijie Guo: @Rodrigo Malacarne: 2.2 is released. I
think c++ and python support compaction topic now in 2.2.
pray : Rodrigo Malacarne
bananadance : Rodrigo Malacarne
----
2018-10-25 15:22:13 UTC - Matteo Merli: Yes , it was actually out in 2.1.1,
though the doc is outdated
+1 : Rodrigo Malacarne
----
2018-10-25 16:58:33 UTC - Aniket: Hello, Pulsar seems very interesting. While
going through Pulsar documentation I came across geo replication and I have a
question in regards to it.
<http://pulsar.apache.org/docs/latest/admin/GeoReplication/>
If messages are replicated across, lets say three DCs, then can we achieve
`exactly once` semantics across all the clusters in three DCs?
In other words, if consumer in DC1 has consumed the event from the queue, can
we avoid the consumers in DC2 and DC3 processing the same event?
----
2018-10-25 16:59:58 UTC - Matteo Merli: You can achieve exactly-once for each
of them independently, though not by having them share data
----
2018-10-25 17:00:35 UTC - Matteo Merli: @Matteo Merli set the channel topic:
Pulsar graduated as a Top-Level project :tada: - Pulsar release 2.2.0 -
<https://pulsar.apache.org/release-notes/#2.2.0>
tada : Rodrigo Malacarne, Ali Ahmed
----
2018-10-25 17:37:24 UTC - Aniket: @Matteo Merli ok
----
2018-10-25 17:37:27 UTC - Aniket: thanks
----
2018-10-25 17:50:01 UTC - Matteo Merli: @Aniket each subscription is local to a
particular cluster
----
2018-10-25 18:09:39 UTC - Rodrigo Malacarne: @Matteo Merli, is Pulsar SQL
integration already done in 2.2.0? If yes, is there any documentation already
available about how to query topics from Pulsar command line?
----
2018-10-25 18:11:47 UTC - Aniket: @Matteo Merli yeah that makes sense. Probably
if we want some kind of de-duplication then we will have to build some storage
on top of pulsar. I am thinking having cross DC replication helps high
availability, but there can be a case where we don’t to consume the data twice.
----
2018-10-25 18:53:33 UTC - Rodrigo Malacarne: Hello all, I'm kind of stuck on
how to correctly start bookkeeper as a deamon service with Pulsar... when I use
```bin/pulsar-daemon start bookie```
I'm not able to set the number of bookies I want to be used in bookkeeper...
I've searched inside bookkeeper.conf and I couldn't find any parameter related
to the amount of bookies... I'm sure I'm missing something. Could anybody
please advise me on how to set the amount of bookies when starting bookkeeper
as a daemon service?
----
2018-10-25 18:54:11 UTC - Ali Ahmed: ./bin/pulsar-daemon start bookie starts
one bookie on that machine.
----
2018-10-25 18:54:38 UTC - Rodrigo Malacarne: @Ali Ahmed, that's ok .. but what
if I want to start, say, 4 bookies?
----
2018-10-25 18:54:45 UTC - Ali Ahmed: you can do the same on other machines to
bring up more bookies
----
2018-10-25 18:55:27 UTC - Rodrigo Malacarne: @Ali Ahmed, so the maximum number
of bookies per machine on any cluster is limited to 1?
----
2018-10-25 18:55:44 UTC - Ali Ahmed: No
----
2018-10-25 18:56:25 UTC - Ali Ahmed: but bookies are meant to represent a
storage node on a cluster, generally one bookie is mapped to one machine
+1 : Rodrigo Malacarne
----
2018-10-25 18:57:22 UTC - Rodrigo Malacarne: @Ali Ahmed, thank you very much!
----
2018-10-25 19:10:55 UTC - Matteo Merli: bookies are identified by `IP:port`.
You can start multiple bookies by passing different config files to use
different ports
+1 : Rodrigo Malacarne
----
2018-10-25 19:12:40 UTC - Matteo Merli: > is Pulsar SQL integration already
done in 2.2.0? If yes, is there any documentation already available about how
to query topics from Pulsar command line?
The SQL docs were there in the “next” section of the website but appears to be
gone now.. checking
----
2018-10-26 00:35:05 UTC - Mark Poole: @Mark Poole has joined the channel
----
2018-10-26 06:57:06 UTC - Harald Gustafsson: Hi, for my test framework I
constantly create tenant and namespace and then delete them at end of tests.
Using the same name for the namespace in each run of the test. But when I check
it seems that the deleted flag can't be set to false after the namespace have
been deleted once. Is there a way of clearing a namespace so that it is
completely removed so the tests will come to a clean setup each time.
----
2018-10-26 08:23:14 UTC - Xavier Baques: Hi there, quick quesiton: is the
Pulsar 2.2.0 client is 100% compatible with a 2.1.x cluster? Also what is the
client-cluster compatibility policy as it is not clear to me from the release
notes. Thanks!
----
2018-10-26 08:30:17 UTC - xiangzi: @xiangzi has joined the channel
----
