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 2019/10/02 09:11:03 UTC
Slack digest for #general - 2019-10-02
2019-10-01 14:29:52 UTC - Andrey Stelmashenko: @Andrey Stelmashenko has joined the channel
----
2019-10-01 14:44:27 UTC - David Kjerrumgaard: @Gaetan SNL There have been requests for similar functionality from other users, and we are currently working on a proposal to allow subscription pruning. Basically we are considering adding some policies that allow a user to determine when a subscription should be considered "dead" and get removed. Thoughts are focused on how long since a message was published to the subscription, etc. If you have more detailed requirements feel free to provide more input
----
2019-10-01 14:54:20 UTC - Britt Bolen: @Britt Bolen has joined the channel
----
2019-10-01 15:55:16 UTC - Matteo Merli: @Gaetan SNL You can do that with the `Reader` API in PulsarClient
----
2019-10-01 15:55:59 UTC - Matteo Merli: A reader is similar to a consumer, though is ephemeral and it doesn’t imply data retention.
----
2019-10-01 15:56:23 UTC - Matteo Merli: You can set time-retention to ensure the data is kept for a specific amount of time.
----
2019-10-01 17:08:15 UTC - Jesse Zhang (Bose): @Matteo Merli @Sijie Guo do you know if this is a bug?
----
2019-10-01 17:09:45 UTC - Matteo Merli: The delay for the nack redelivery is not meant to be precise, rather a way to have some automatic backoff in the retries. Having said that… it shouldn’t be 1/4th of the specified value either.
----
2019-10-01 17:25:44 UTC - Jesse Zhang (Bose): Agree! Given I can consistently reproduce the issue, please let me know if this is confirmed to be a bug later. Currently, i have set 3* `my expected delay time` to NackRedeliveryDelay as a work around, seems working, but please let me know if this has other side-effects.
----
2019-10-01 17:40:21 UTC - Matteo Merli: No, there shouldn’t be any side effect in increasing that
+1 : Jesse Zhang (Bose)
----
2019-10-01 18:10:41 UTC - Luke Lu: We’re seeing this (topic unavailable due bundle being unloaded) in 2.4.1:
```
"00:52:30.838 [pulsar-io-24-2] INFO org.apache.pulsar.broker.lookup.TopicLookupBase - Failed to lookup null for topic persistent://... with error Namespace bundle .../0x00000000_0x40000000 is being unloaded\n" "00:52:30.838 [pulsar-io-24-2] INFO org.apache.pulsar.broker.service.ServerCnx - [/10.244.0.50:47136] Subscribing on topic persistent://...\n" "00:52:30.838 [pulsar-io-24-2] WARN org.apache.pulsar.broker.service.persistent.PersistentTopic - [persistent://...] Attempting to subscribe to a fenced topic\n" "00:52:30.838 [pulsar-io-24-2] WARN org.apache.pulsar.broker.service.ServerCnx - [/10.244.0.50:47136][persistent://...][216569ee-9823-4266-898e-1c54001dfa05] Failed to create consumer: Topic is temporarily unavailable\n" "java.util.concurrent.CompletionException: org.apache.pulsar.broker.service.BrokerServiceException$TopicFencedException: Topic is temporarily unavailable\n" "u0009at java.util.concurrent.CompletableFuture.encodeRelay(CompletableFuture.java:326) ~[?:1.8.0_212]\n" "u0009at java.util.concurrent.CompletableFuture.uniComposeStage(CompletableFuture.java:984) ~[?:1.8.0_212]\n" "u0009at java.util.concurrent.CompletableFuture.thenCompose(CompletableFuture.java:2124) ~[?:1.8.0_212]\n" "u0009at org.apache.pulsar.broker.service.ServerCnx.lambda$null$13(ServerCnx.java:680) ~[org.apache.pulsar-pulsar-broker-2.4.1.jar:2.4.1]\n" "u0009at java.util.concurrent.CompletableFuture.uniApply(CompletableFuture.java:602) ~[?:1.8.0_212]\n" "u0009at java.util.concurrent.CompletableFuture.uniApplyStage(CompletableFuture.java:614) ~[?:1.8.0_212]\n"
```
Which appeared to be causing producer timeouts. This is supposed to be fixed in 2.2.1: <https://github.com/apache/pulsar/issues/2967>
----
2019-10-02 06:29:29 UTC - Jeff: @Jeff has joined the channel
----
2019-10-02 06:37:56 UTC - Jeff: Just wondering if anyone has released a Heron Streamlet source for Pulsar? The alternative would be to take the storm spout, patch it up, and wrap it into a streamlet. But I thought there must be one already out there that Google just isn't turning up...
----
2019-10-02 06:41:08 UTC - Ali Ahmed: @Jeff here you go <https://github.com/apache/incubator-heron/blob/5ee2490c025c7684c097962cf6d25f54ead73cf4/examples/src/java/org/apache/heron/examples/streamlet/SimplePulsarSourceTopology.java>
----
2019-10-02 06:41:34 UTC - Ali Ahmed: this can be enhanced further with pulsar’s types safe api
----
2019-10-02 06:43:12 UTC - Jeff: Oh excellent! Glad I asked, thanks very much
----
2019-10-02 07:32:06 UTC - Ali Ahmed: I have created a vcpkg port for installing libpulsar on windows
<https://github.com/aahmed-se/vcpkg>
here are the instructions for windows machines
```
git clone <https://github.com/aahmed-se/vcpkg>
cd vcpkg
bootstrap-vcpkg.bat
vcpkg.exe install libpulsar:x64-windows
```
----