Displaying #cassandra-dev/2017-02-13.log:

Mon Feb 13 01:15:34 2017  Vijay_:Joined the channel
Mon Feb 13 01:21:09 2017  kohlisankalp:Joined the channel
Mon Feb 13 02:17:43 2017  kohlisankalp:Joined the channel
Mon Feb 13 03:14:28 2017  jasobrown:Joined the channel
Mon Feb 13 04:27:08 2017  ynazarov:Joined the channel
Mon Feb 13 07:41:44 2017  makrusak:Joined the channel
Mon Feb 13 08:03:45 2017  spodkowinski:Joined the channel
Mon Feb 13 08:42:50 2017  kvaster:Joined the channel
Mon Feb 13 08:51:12 2017  RussSpitzer:Joined the channel
Mon Feb 13 09:10:23 2017  minimarcel_:Joined the channel
Mon Feb 13 09:26:13 2017  iksaif:hello ptnapoleon, aweisberg told me I should ask you to get access to cassci :) (fork is https://github.com/iksaif/cassandra) Followed http://cassci.datastax.com/userContent/cassci-usage.html but I don't have access to the mentioned google docs
Mon Feb 13 10:10:58 2017  kvaster:hi. can anybody comment about CompactionStrategyManager.getCompactionStrategyIndex ? It is called for each sstable during startup. And this function calls StorageService.getDiskBoundaries which is really slow in some cases.... Can we cache result of StorageService.getDiskBoundaries ?
Mon Feb 13 10:27:15 2017  minimarcel_:Joined the channel
Mon Feb 13 12:05:46 2017  krolser:Joined the channel
Mon Feb 13 12:07:12 2017  krolser:Left the channel
Mon Feb 13 12:10:15 2017  krolser:Joined the channel
Mon Feb 13 12:51:23 2017  marcuse:kvaster: we can, could you create a ticket?
Mon Feb 13 12:52:43 2017  kvaster:Yes, I will. It seems that getCompactionStrategyIndex can be called not only during startup -> and it can be really slow. I've tried to implement simple cache and that reduced node startup time from 25minutes to one minute.
Mon Feb 13 12:57:00 2017  spodkowinski:since the disk boundaries calculation depends on the local ranges, I'm wondering what happens in case of changed ranges? will the same partitions still get mapped to identical directories afterwards?
Mon Feb 13 12:58:20 2017  kvaster:I've tried with cache which depends on both ColumnFamily and Directories
Mon Feb 13 12:59:20 2017  kvaster:Anyway getDiskBoundaries is a big problem at our setup.
Mon Feb 13 13:03:14 2017  makrusak:Joined the channel
Mon Feb 13 13:05:05 2017  marcuse:we can just listen for ring changes and invalidate the cache then
Mon Feb 13 13:24:13 2017  kvaster:I've created 'bug report' - https://issues.apache.org/jira/browse/CASSANDRA-13215
Mon Feb 13 13:24:50 2017  kvaster:My 'solution' is too simple and I don't think it's really good. Invalidating cache for ring changes is a better idea.
Mon Feb 13 13:30:38 2017  driftx_:Joined the channel
Mon Feb 13 13:30:45 2017  lqid:Joined the channel
Mon Feb 13 13:31:12 2017  exlt_:Joined the channel
Mon Feb 13 14:28:18 2017  ptnapoleon:iksaif: Sorry, but I've got bad news for you then. We're in the process of transitioning CI away from cassci over to ASF run services, and part of that is to stop onboarding people to cassci.
Mon Feb 13 14:38:12 2017  clohfink:Joined the channel
Mon Feb 13 14:38:32 2017  clohfink:Joined the channel
Mon Feb 13 16:05:48 2017  iksaif:ptnapoleon: ok, well I guess I'll wait for that and ask people who have access to run stuff
Mon Feb 13 16:06:00 2017  ynazarov:Joined the channel
Mon Feb 13 16:10:32 2017  ptnapoleon:iksaif: It's my understanding that it will be totally ready this week or next, and that committers and/or PMC will have access
Mon Feb 13 16:12:55 2017  iksaif:ok cool !
Mon Feb 13 16:16:12 2017  exlt_:alternatively, if you'd like help setting up your own internal CI, I can help with that
Mon Feb 13 16:18:20 2017  iksaif:exlt: I work on a laptop with limited RAM so I'll wait for now, but we may set that up later if getting access to the community one is too troublesome
Mon Feb 13 16:20:43 2017  aweisberg:exlt: That's an interesting proposition. I have two boxes I could use as slaves if I were to burn some watts.
Mon Feb 13 16:20:47 2017  joaquin:Joined the channel
Mon Feb 13 16:23:16 2017  exlt:well, in order to run jobs on the ASF Jenkins, it requrires one to be a committer and be added to the 'hudson-jobadmin' LDAP group - https://cwiki.apache.org/confluence/display/INFRA/Jenkins - someone who's not a committer will indeed need someone to run jobs on their behalf
Mon Feb 13 16:24:08 2017  exlt:I'll be sending email to -dev and such, once the details are worked out - sorry to just be dropping random tidbits on irc..
Mon Feb 13 16:25:01 2017  exlt:reviewers will probably be key in getting jobs run on ASF's jenkins
Mon Feb 13 16:28:05 2017  exlt:aweisberg: I'll also reiterate adding resources to ASF infra - there's a bit of ongoing process discussion with INFRA on adding project servers to the ASF Jenkins and it's working OK, with a few little hurdles
Mon Feb 13 16:44:01 2017  kohlisankalp:Joined the channel
Mon Feb 13 17:02:48 2017  aweisberg:So I am looking at the instructions on how to commit http://cassandra.apache.org/doc/latest/development/how_to_commit.html and it looks like they result in multiple versions of each commit if you follow the branch based approach since you merge in the previous branch using -s ours and then add the changes on top? I didn't realize we were doing that?
Mon Feb 13 17:03:24 2017  tjake:yup
Mon Feb 13 17:08:55 2017  exlt:that's if it's not a clean merge, depending on branch/patch deltas
Mon Feb 13 17:17:20 2017  Vijay:Joined the channel
Mon Feb 13 17:33:17 2017  mstepura:Joined the channel
Mon Feb 13 17:36:28 2017  krolser:Joined the channel
Mon Feb 13 17:42:49 2017  krolser:Joined the channel
Mon Feb 13 18:00:09 2017  JayZhuang:Joined the channel
Mon Feb 13 18:05:46 2017  kvaster:Joined the channel
Mon Feb 13 18:15:38 2017  kohlisan_:Joined the channel
Mon Feb 13 18:29:57 2017  kohlisankalp:Joined the channel
Mon Feb 13 18:35:59 2017  aweisberg:I had gotten it wrong. You are amending the merge commit to contain the correct stuff. So you don't end up with any extra commits. Never though of amending the merge commit before. That's clever.
Mon Feb 13 18:51:50 2017  jeffj:the how-to-commit docs leave out 'Patch by X, Reviewed by Y for CASSANDRA-Z' which is - presumably- expected at this point
Mon Feb 13 18:52:12 2017  jeffj:lot of new comitters, should have that in writing
Mon Feb 13 18:52:30 2017  aweisberg:Yeah Jason had to point it out to me.
Mon Feb 13 18:52:33 2017  jeffj:(i know i missed it on the first one i did)
Mon Feb 13 18:55:24 2017  aweisberg:No matter I can't figure out how to push anyways :-P
Mon Feb 13 18:55:31 2017  aweisberg:https://www.irccloud.com/pastebin/XZTNmnJV/
Mon Feb 13 18:56:06 2017  aweisberg:Jason is looking at it.
Mon Feb 13 18:57:13 2017  tjake:aweisberg: I use http://git-wip-us.apache.org/repos/asf/cassandra.git
Mon Feb 13 18:57:38 2017  jeffj:https:// is the writable endpoint
Mon Feb 13 18:57:42 2017  jeffj:http:// is read-only
Mon Feb 13 18:58:15 2017  exlt:git.apache.org is similar to github - it's just a mirror
Mon Feb 13 18:59:51 2017  jeffj:or maybe http:// works for writes, too
Mon Feb 13 19:01:17 2017  aweisberg:http worked although it requires me to auth. I'll try it as a git url next time.
Mon Feb 13 19:01:21 2017  aweisberg:tjake: thanks!
Mon Feb 13 19:01:48 2017  exlt:aweisberg: https://git.apache.org did?
Mon Feb 13 19:02:10 2017  aweisberg:Jake's URL http://git-wip-us.apache.org/repos/asf/cassandra.git
Mon Feb 13 19:02:19 2017  tjake:I like the auth
Mon Feb 13 19:02:39 2017  exlt:yep yep, thanks - I thought the mirror took a commit :)
Mon Feb 13 19:02:48 2017  exlt:that would be bad
Mon Feb 13 19:02:51 2017  tjake:helps avoid accidental pushes
Mon Feb 13 19:02:55 2017  aboudreault:Joined the channel
Mon Feb 13 19:02:55 2017  aboudreault:Joined the channel
Mon Feb 13 19:07:01 2017  aweisberg:tjake: That's a good idea I'll stick with the auth
Mon Feb 13 19:07:21 2017  tjake:aweisberg: congrats btw!
Mon Feb 13 19:11:24 2017  aweisberg:tjake: Thanks again. Talk to me again after we find out if my first commit was up to code =)
Mon Feb 13 19:17:35 2017  driftx:pun intended? heh
Mon Feb 13 19:17:59 2017  JayZhuang:Joined the channel
Mon Feb 13 20:10:53 2017  mstepura:Joined the channel
Mon Feb 13 20:22:23 2017  aweisberg:driftx: Pun certainly achieved
Mon Feb 13 20:23:48 2017  dikang:Joined the channel
Mon Feb 13 20:54:33 2017  JayZhuang:Joined the channel
Mon Feb 13 21:02:16 2017  kohlisankalp:Joined the channel
Mon Feb 13 21:09:38 2017  kohlisankalp:Joined the channel
Mon Feb 13 21:19:29 2017  bbromhead:Joined the channel
Mon Feb 13 21:23:17 2017  mstepura:Joined the channel
Mon Feb 13 21:27:34 2017  bdeggleston:is the trunk_offheap_dtest cassci job using the same dtest branch as trunk_dtest?
Mon Feb 13 21:28:01 2017  bdeggleston:is has a bunch of failures that should be fixed in dtest master, and that aren't failing in trunk_dtest
Mon Feb 13 21:28:09 2017  minimarcel_:Joined the channel
Mon Feb 13 21:30:32 2017  makrusak:Joined the channel
Mon Feb 13 21:31:19 2017  ptnapoleon:It is using hte same branch
Mon Feb 13 21:31:27 2017  ptnapoleon:It runs less often
Mon Feb 13 21:31:34 2017  ptnapoleon:and, obviously, offheap memtables are used
Mon Feb 13 21:31:37 2017  ptnapoleon:Those are the only differences
Mon Feb 13 21:32:18 2017  bdeggleston:ah got it thanks
Mon Feb 13 22:01:00 2017  kohlisankalp:Joined the channel
Mon Feb 13 22:07:24 2017  mstepura:Joined the channel
Mon Feb 13 22:43:48 2017  bbromhead:Joined the channel
Mon Feb 13 22:46:05 2017  bbromhead:Joined the channel
Mon Feb 13 22:55:52 2017  bbromhead:Joined the channel
Mon Feb 13 23:03:42 2017  dikang:Joined the channel
Mon Feb 13 23:29:32 2017  minimarcel_:Joined the channel

Comments