#general
@manish.sharma1: @manish.sharma1 has joined the channel
@nair.a: Hi Team, Few questions regarding Realtime Table. Cluster setup: 2 servers, 1 controller, 1 Broker. 1) If one of the server is down, consumption is completely stopped. why other server doesn't start consuming? 2) if consumer.prop.auto.offset.reset is set to smallest, from which offset it will start consuming? from last committed offset in zk? 3) If i change table config , does it reflect straight away ? Thanks.
@mayanks: 1. Servers are assigned partitions to consume from. If a server goes down it’s partitions are not auto moved to other server. This is why you need to run in HA mode (relocation).
@nair.a: Hey @mayanks , thanks for responding. By HA mode, you mean replication?
@mayanks: Yes. Replicas will consume the same partitions. So if one goes down, other still consume
@mayanks: 2. For new table consuming for the first time, it will start from smallest Kafka offset. If consumption already started then offset in zk used
@mayanks: 3. Depends on the change you made to config. Some require reload api (indexing changes), others wait until new segment consumption starts etc (streaming config changes).
@nair.a: Any documentation available ?
@mayanks: What changes are you making? We should add documentation if missing cc: @mark.needham
@nair.a: I changed the offset reset from latest to smallest.
@ssubrama: changing the offset to latest will not affect anything other than table creation. if you drop the table and re-create it, then it will start consuning from latest (but if you do this, either use a new table name, or let the old table drain completely in terms of externalview, thanks)
@matteobovetti: @matteobovetti has joined the channel
@arun.rajan.work: @arun.rajan.work has joined the channel
#random
@manish.sharma1: @manish.sharma1 has joined the channel
@matteobovetti: @matteobovetti has joined the channel
@arun.rajan.work: @arun.rajan.work has joined the channel
#troubleshooting
@manish.sharma1: @manish.sharma1 has joined the channel
@yeongjukang: Hi, I have some questions related to indices. 1. Is Forward index applied to all non specified columns of a table in default? 2. Are there ways to see query execution plans including index usage? I tried from explain, explain (type distributed), explain (type io) from presto but failed to find useful information from that 3. Some index files doesn’t seem to be purged after table deletion. Should I delete those myself if I have to make table with same name? (I didn’t try re-generation of the behavior) Thank you for your effort to such a nice software!
@xiangfu0: 1. yes, it’s pinot’s data representation 2. it’s not exposed yet, we are working on this support, ref:
@yeongjukang: @xiangfu0 Thanks!
@yeongjukang: Hi, I have 5 servers(v0.9) in my cluster and one of them turned to dead state. server’s process never printed crash log and last lines of the server log looks like its state seems okay. but table state at zookeeper turned to offline and i cannot see my node from liveinstances. I am running my server on EKS and it never had pod restart. Is there anything I can do before pod restart?
@xiangfu0: Did your set resources section for helmChart?
@xiangfu0: You can make request and limit to be same for memory and cpu
@xiangfu0: Also check JAVA_OPTS and give a reasonable xmx
@yeongjukang: @xiangfu0 thanks, i will try after this case. My cluster could have resource race condition. But there was no podeviction. By the way is there anything i can do before helm upgrade?
@alihaydar.atil: Hello Everyone, is it possible to change H3 index resolution after table creation?
@mark.needham: I think you should be able to do it like this:
@alihaydar.atil: @mark.needham Thank you for the detailed presentation!
@matteobovetti: @matteobovetti has joined the channel
@arun.rajan.work: @arun.rajan.work has joined the channel
@deemish2: Hello everyone , i am not able to start zookeeper using pinot-0.9.0 with command - bin/pinot-admin.sh StartZookeeper . Please help
@bagi.priyank: add ```-Dpinot.admin.system.exit=false``` to JAVA_OPTS
@bagi.priyank: you will have to do this for other components as well
@bagi.priyank: more info in this thread
#pinot-dev
@ken: Can the master branch still be built using Java 1.8? Just pulled and tried `mvn clean install -DskipTests -Pbin-dist -T 4 -Djdk.version=8` and got a strange error: ```[ERROR] pinot/pinot-common/src/test/java/org/apache/pinot/common/metrics/PinotMetricUtilsTest.java:[43,88] error: cannot access PinotMetricsRegistry [ERROR] class file for PinotMetricsRegistry not found [ERROR] pinot/pinot-common/src/test/java/org/apache/pinot/common/metrics/PinotMetricUtilsTest.java:[77,76] error: incompatible types: PinotMetricsRegistry cannot be converted to org.apache.pinot.spi.metrics.PinotMetricsRegistry [ERROR] pinot/pinot-common/src/test/java/org/apache/pinot/common/metrics/PinotMetricUtilsTest.java:[78,36] error: incompatible types: org.apache.pinot.spi.metrics.PinotMetricsRegistry cannot be converted to PinotMetricsRegistry```
@xiangfu0: if CI passed then should be. Are you on openJDK 8?
#getting-started
@matteobovetti: @matteobovetti has joined the channel
#releases
@matteobovetti: @matteobovetti has joined the channel
--------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@pinot.apache.org For additional commands, e-mail: dev-h...@pinot.apache.org
