2020-04-14 15:11:04 UTC - Raman Gupta: There are
<https://github.com/apache/pulsar/wiki/PIP-43:-producer-send-message-with-different-schema|two>
<https://github.com/apache/pulsar/wiki/PIP-43:-Separate-schema-compatibility-checker-for-producer-and-consumer|different>
PIP-43s... presumably
<https://github.com/apache/pulsar/wiki/PIP-43:-Separate-schema-compatibility-checker-for-producer-and-consumer|this
one> is out of date and should be deleted or renamed?
----
2020-04-14 16:01:44 UTC - Addison Higham: Yes, I do believe it is likely
because my background threads aren't dying, but this isn't an external signal,
this is an exception from within the JVM that bubbles all the way up to
`JavaInstanceStarter` killing the function run loop. At some point, at least in
the mode where their is only a single function per JVM, it would make sense to
just forcefully close the JVM (perhaps with a timeout)
----
2020-04-14 16:02:35 UTC - Addison Higham: (sorry got distracted from this convo
yesterday)
----
2020-04-14 16:08:31 UTC - Sijie Guo: One is PIP-44
----
2020-04-14 16:08:34 UTC - Sijie Guo: Let me fix it
----
2020-04-14 16:09:33 UTC - Raman Gupta: Ah, there is already a PIP-44 page, so
<https://github.com/apache/pulsar/wiki/PIP-43:-Separate-schema-compatibility-checker-for-producer-and-consumer>
should be deleted.
----
2020-04-14 16:11:24 UTC - matt_innerspace.io: Using `pulsar 2.4.0` (because
WindowFunctions don't work in later versions, waiting for 2.5.1 release which
fixes it) i see an error generate from a function's debug statement and i can't
see where the LOG_LEVEL is set to debug anywhere. I figure if i set the
log_level to INFO, it'll skip this log line and the error will go away.
```2020-04-09 10:38:24,988 tenant/namespace/function-0 WARN
org.apache.logging.slf4j.Log4jLogger caught java.lang.NullPointerException
logging ParameterizedMessage: [{}] [{}] add message to batch, num messages in
batch so far {} java.lang.NullPointerException```
which appears to happen in this class/version -
<https://github.com/apache/pulsar/blob/b45736ad0738116c9c3cae27ed18f1342b55139e/pulsar-client/src/main/java/org/apache/pulsar/client/impl/BatchMessageContainerImpl.java>
pulsar standalone's `conf/log4j2.yaml` appears to set everything to `info`
level.
what am i missing?
----
2020-04-14 16:14:35 UTC - Sijie Guo: I seems that it is a problem in 2.4.0.
----
2020-04-14 16:14:54 UTC - Sijie Guo: why not use 2.4.2?
----
2020-04-14 16:15:52 UTC - matt_innerspace.io: the `WindowFunction` fix is in
2.4.3 or 2.5.1 - <https://github.com/apache/pulsar/pull/6246>
----
2020-04-14 16:16:33 UTC - matt_innerspace.io: i can't register a window
function via pulsar cli in anything later than 2.4.0, unless there is another
way to do it.
----
2020-04-14 16:16:35 UTC - Sijie Guo: okay. I was confused when you say `pulsar
2.4.0`
----
2020-04-14 16:16:49 UTC - Sijie Guo: I see
----
2020-04-14 16:16:53 UTC - Sijie Guo: nvm
----
2020-04-14 16:18:06 UTC - Sijie Guo: since 2.5.1 is under voting, maybe you can
try the release candidate 3 -
<https://dist.apache.org/repos/dist/dev/pulsar/pulsar-2.5.1-candidate-3/>
----
2020-04-14 16:23:15 UTC - Addison Higham: @Sijie Guo on that note of RC3, are
there docker images for RCs pushed anywhere?
100 : Konstantinos Papalias, matt_innerspace.io
eyes : Konstantinos Papalias
----
2020-04-14 16:48:48 UTC - Sijie Guo: good point. in the current process, the
docker image was generated from a release tag after it passes voting. we need
to change the release process a bit. It should be doable.
pray : Konstantinos Papalias
----
2020-04-14 16:49:58 UTC - Sijie Guo: @tuteng Can you create a docker image for
release candidates so that people can verify using the docker image of a
release candidate tag?
+1 : matt_innerspace.io
----
2020-04-14 16:50:51 UTC - Konstantinos Papalias: that would be really valuable
----
2020-04-14 16:52:49 UTC - Addison Higham: I was just about to go deploy RC3 but
then realized I need to build it first. Going to do that anyway, but in the
future would probably help with release testing
+1 : matt_innerspace.io
----
2020-04-14 16:55:19 UTC - Sijie Guo: yes totally agreed.
----
2020-04-14 17:26:26 UTC - Addison Higham: one more note on build: I had to
manually delete
`pulsar-client/src/test/java/org/apache/pulsar/client/avro/generated` folder.
`mvn clean` did not remove it
+1 : Konstantinos Papalias
----
2020-04-14 17:26:50 UTC - Matteo Merli: Yes, it's a problem with the Avro maven
plugin
----
2020-04-14 17:27:04 UTC - Matteo Merli: all the generated stuff should be put
under `target/...`
----
2020-04-14 17:34:38 UTC - Matteo Merli: Not sure if there's an easy fix or
configuration option
----
2020-04-14 19:10:50 UTC - Addison Higham: one more build issue: it looks a
`docker build` is happening with a `--pull` option. Had an issue with not being
able to find a package because I had an old base image that was used
----
2020-04-14 19:23:19 UTC - matt_innerspace.io: docker images for release
candidates would be great!
----
2020-04-14 20:45:01 UTC - Addison Higham: just finishing uploading
`instructure/pulsar-all:2.5.1-rc3` if anyone wants to try it out
+1 : Sijie Guo
----
2020-04-14 20:45:36 UTC - Addison Higham: (will be a few more minutes, upload
bandwidth from home isn't great, the biggest downside of WFH!)
----
