Displaying #cassandra-dev/2017-05-22.log:

Mon May 22 00:00:23 2017  clohfink:Joined the channel
Mon May 22 00:03:33 2017  clohfink:Joined the channel
Mon May 22 00:45:44 2017  clohfink:Joined the channel
Mon May 22 01:34:56 2017  tolbertam:Joined the channel
Mon May 22 01:39:39 2017  clohfink:Joined the channel
Mon May 22 02:44:53 2017  clohfink:Joined the channel
Mon May 22 03:00:00 2017  clohfink:Joined the channel
Mon May 22 03:07:06 2017  Vijay_:Joined the channel
Mon May 22 03:31:10 2017  Vijay_:Joined the channel
Mon May 22 03:57:51 2017  clohfink:Joined the channel
Mon May 22 05:10:45 2017  clohfink:Joined the channel
Mon May 22 05:34:11 2017  tolbertam:Joined the channel
Mon May 22 06:23:44 2017  cosql:Joined the channel
Mon May 22 06:32:06 2017  spodkowinski:Joined the channel
Mon May 22 06:59:20 2017  clohfink:Joined the channel
Mon May 22 07:05:44 2017  cosql:Joined the channel
Mon May 22 07:06:32 2017  cosql:Joined the channel
Mon May 22 07:34:27 2017  tolbertam:Joined the channel
Mon May 22 08:45:06 2017  kvaster:Joined the channel
Mon May 22 08:47:26 2017  clohfink:Joined the channel
Mon May 22 09:34:44 2017  tolbertam:Joined the channel
Mon May 22 09:48:40 2017  makrusak:Joined the channel
Mon May 22 10:35:43 2017  clohfink:Joined the channel
Mon May 22 11:34:59 2017  tolbertam:Joined the channel
Mon May 22 11:54:59 2017  jmckenzie:Joined the channel
Mon May 22 12:56:43 2017  tjake:urandom: blerer just sent a note to dev about it
Mon May 22 13:08:28 2017  blerer:marcuse: Do you know how much we rely on SSTableReader:readCount for SizedTierd compaction? As it is only increase for single partition queries I was wondering how much problem it would cause if the users were doing a lot of partition range queries
Mon May 22 13:11:14 2017  marcuse:blerer: not sure how much it helps in real life
Mon May 22 13:13:20 2017  marcuse:what we do is we prioritize "hot" sstables when picking candidates, so maybe by not updating the meters on range queries we might just not compact the sstables as quickly as we should
Mon May 22 13:13:26 2017  marcuse:but doubt it is a real problem
Mon May 22 13:14:11 2017  blerer:Great, thanks.
Mon May 22 13:18:06 2017  clohfink:Joined the channel
Mon May 22 13:34:10 2017  tolbertam:Joined the channel
Mon May 22 14:20:08 2017  clohfink:Joined the channel
Mon May 22 14:21:21 2017  spodkowinski:did anyone successfully compile trunk against java 9 yet?
Mon May 22 14:22:12 2017  spodkowinski:haven't had any luck to "unlock" any of the internal api's again that are not accessible with jdk9 anymore
Mon May 22 14:22:57 2017  iamaleksey:snazy may have
Mon May 22 14:23:33 2017  iamaleksey:I don't foresee us supporting Java 9 for a long time however; nor do I think we should care (or that we'll be unique about it)
Mon May 22 14:24:44 2017  clohfink:shenandoah might be worth it
Mon May 22 14:25:07 2017  iamaleksey:it would've been if it was destined for Java 9
Mon May 22 14:25:14 2017  iamaleksey:or production ready in any way
Mon May 22 14:25:26 2017  clohfink:oh its not gonna be in 9?
Mon May 22 14:26:14 2017  iamaleksey:no
Mon May 22 14:26:28 2017  iamaleksey:might never make it to JDK proper ever
Mon May 22 14:26:48 2017  iamaleksey:I would expect Java 11 earliest if at all, myself
Mon May 22 14:27:45 2017  spodkowinski:keep in mind java 8 will be EOL not too long after 9 (I think 1 year?).. at some point we want to be able to at least enable users to run cassandra on 9 jre
Mon May 22 14:51:13 2017  clohfink:Joined the channel
Mon May 22 15:01:16 2017  jeffj:i tried shenandoah with jdk8 using shipilev's backport and saw no real difference on a toy ccm cluster. but toy+jdk8backport, who knows.
Mon May 22 15:09:10 2017  EdwardCapriolo:Joined the channel
Mon May 22 15:34:25 2017  tolbertam:Joined the channel
Mon May 22 15:52:59 2017  urandom:tjake, blerer: thanks, is there a ticket i can reference?
Mon May 22 15:54:22 2017  urandom:oh, #13120
Mon May 22 15:54:22 2017  CassBotJr:https://issues.apache.org/jira/browse/CASSANDRA-13120 (Open; Unresolved; Unscheduled): "Trace and Histogram output misleading"
Mon May 22 15:54:25 2017  urandom:duh.
Mon May 22 16:38:50 2017  kohlisankalp:Joined the channel
Mon May 22 16:55:30 2017  clohfink:Joined the channel
Mon May 22 17:12:42 2017  mebigfatguy:Joined the channel
Mon May 22 17:13:40 2017  cosql:Joined the channel
Mon May 22 18:05:52 2017  zaller:Joined the channel
Mon May 22 18:17:19 2017  clohfink:Joined the channel
Mon May 22 18:35:10 2017  makrusak:Joined the channel
Mon May 22 18:45:35 2017  clohfink:Joined the channel
Mon May 22 18:48:59 2017  kvaster:Joined the channel
Mon May 22 19:09:52 2017  aweisberg:Can anyone think of a reason why if we supported StartTLS for the intra-cluster connections we would still need to support having separate SSL and non-SSL ports? Separate ports would basically become an unsupported configuration in 4.0 if you intend to run multiple instances on the same network interface in 4.0 (currently you can't do it all, but still).
Mon May 22 19:09:52 2017  aweisberg:Existing configurations would continue to work fine.
Mon May 22 19:19:14 2017  clohfink:Joined the channel
Mon May 22 19:30:19 2017  aboudreault:Joined the channel
Mon May 22 19:58:24 2017  mebigfatguy:Joined the channel
Mon May 22 20:32:41 2017  nickmbailey:Joined the channel
Mon May 22 20:42:19 2017  aboudreault:Joined the channel
Mon May 22 21:26:23 2017  nickmbailey:Joined the channel
Mon May 22 21:30:19 2017  aboudreault:Joined the channel
Mon May 22 21:55:11 2017  nickmbailey:Joined the channel
Mon May 22 21:59:19 2017  aboudreault:Joined the channel
Mon May 22 22:04:40 2017  nickmbailey:Joined the channel
Mon May 22 22:09:35 2017  nickmbailey:Joined the channel
Mon May 22 22:09:55 2017  nickmbailey:Joined the channel
Mon May 22 22:20:23 2017  nickmbailey:Joined the channel
Mon May 22 22:53:45 2017  nickmbailey:Joined the channel
Mon May 22 23:09:20 2017  zaller:Joined the channel
Mon May 22 23:49:50 2017  aboudreault:Joined the channel

Comments