You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "Bob Paulin (Jira)" <ji...@apache.org> on 2024/02/21 13:54:00 UTC
[jira] [Comment Edited] (NIFI-11859) Nifi in standalone mode is not able to enable EmbeddedHazelcastCacheManager
[ https://issues.apache.org/jira/browse/NIFI-11859?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17819071#comment-17819071 ]
Bob Paulin edited comment on NIFI-11859 at 2/21/24 1:53 PM:
------------------------------------------------------------
Looks like Hazelcast will enable the MulticastJoiner[1] if autodetect is enabled. And it is by default [2]. This causes Multicast to be enabled when this is run locally
{code:java}
2024-02-20 19:14:41,547 INFO [Timer-Driven Process Thread-6] com.hazelcast.instance.impl.Node [192.168.56.1]:5701 [nifi] [5.3.5] Using Multicast discovery {code}
By disabling the auto detect this does not happen
[https://github.com/apache/nifi/pull/8440]
I get the following in the logs which I believe is expected.
{code:java}
2024-02-20 19:23:15,870 WARN [Timer-Driven Process Thread-2] com.hazelcast.instance.impl.Node [192.168.56.1]:5701 [nifi] [5.3.5] No join method is enabled! Starting standalone. {code}
This should also address the issue above since it appears to be hitting a usecase where the DiscoveryJoiner is enabled. This should also be suppressed by autodetect being disabled [3]. This appears to happen in both 1.x and 2.x
[1] [https://github.com/hazelcast/hazelcast/blob/a3ae01dcbfa32e3b314047506dadf837d54e8e2a/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L946|https://github.com/hazelcast/hazelcast/blob/e3dd651a78e97c6702ce4260e6263d4818fb29b1/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L974]
[2] [https://github.com/hazelcast/hazelcast/blob/e3dd651a78e97c6702ce4260e6263d4818fb29b1/hazelcast/src/main/java/com/hazelcast/config/AutoDetectionConfig.java#L28]
[3] https://github.com/hazelcast/hazelcast/blob/a3ae01dcbfa32e3b314047506dadf837d54e8e2a/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L938
was (Author: bob):
Looks like Hazelcast will enable the Multicast Joinger[1] if autodetect is enabled. And it is by default [2]. This causes Multicast to be enabled when this is run locally
{code:java}
2024-02-20 19:14:41,547 INFO [Timer-Driven Process Thread-6] com.hazelcast.instance.impl.Node [192.168.56.1]:5701 [nifi] [5.3.5] Using Multicast discovery {code}
By disabling the auto detect this does not happen
[https://github.com/apache/nifi/pull/8440]
I get the following in the logs which I believe is expected.
{code:java}
2024-02-20 19:23:15,870 WARN [Timer-Driven Process Thread-2] com.hazelcast.instance.impl.Node [192.168.56.1]:5701 [nifi] [5.3.5] No join method is enabled! Starting standalone. {code}
This should also address the issue above since it appears to be hitting a usecase where the DiscoveryJoiner is enabled. This should also be suppressed by autodetect being disabled [3]. This appears to happen in both 1.x and 2.x
[1] [https://github.com/hazelcast/hazelcast/blob/a3ae01dcbfa32e3b314047506dadf837d54e8e2a/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L946|https://github.com/hazelcast/hazelcast/blob/e3dd651a78e97c6702ce4260e6263d4818fb29b1/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L974]
[2] [https://github.com/hazelcast/hazelcast/blob/e3dd651a78e97c6702ce4260e6263d4818fb29b1/hazelcast/src/main/java/com/hazelcast/config/AutoDetectionConfig.java#L28]
[3]https://github.com/hazelcast/hazelcast/blob/a3ae01dcbfa32e3b314047506dadf837d54e8e2a/hazelcast/src/main/java/com/hazelcast/instance/impl/Node.java#L938
> Nifi in standalone mode is not able to enable EmbeddedHazelcastCacheManager
> ----------------------------------------------------------------------------
>
> Key: NIFI-11859
> URL: https://issues.apache.org/jira/browse/NIFI-11859
> Project: Apache NiFi
> Issue Type: Bug
> Components: Configuration Management
> Affects Versions: 1.22.0
> Reporter: Jeetendra G Vasisht
> Priority: Blocker
> Attachments: embeddedHazelcastNifiControllerservice.PNG, nifi--app.log
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> EmbeddedHazelcastCacheManager Controller gets enabled in cluster mode with "All Nodes" clustering strategy, but having issue when tried to run in standalone mode with "None" clustering strategy. This is observed in Kubernetes Environment and this is coming as part of internal Nifi packaging and any external dependency or code related to Hazelcast is not being used.
> Controller gets stuck in Enabling state:
> !embeddedHazelcastNifiControllerservice.PNG|width=662,height=131!
> Respective Logs have been attached
--
This message was sent by Atlassian Jira
(v8.20.10#820010)