Github user asfgit closed the pull request at:
https://github.com/apache/storm/pull/810
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabl
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150730065
@kishorvpatil Amazing! +1. Test failure is unrelated. (Kafka)
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub a
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150729673
@HeartSaVioR
- Modified to make "nimbus.reassign" unavailable for users.
- Created #815 to decommission variable for 0.10.x branch.
---
If your project
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150712631
@kishorvpatil
No, just marking ```nimbus.reassign``` option as deprecate and leave proper
reason from 0.10.x would be fine.
---
If your project is set up for it
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150711148
@HeartSaVioR I complete agree with decommissioning this configuration used
only for testing. It should be unavailable for users. Let me deprecate this
variable and a
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150705382
@kishorvpatil
With this change users have no option to turn off reassign since it makes
cluster unusable (cluster cannot run topology unless user calls other oper
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150703213
@HeartSaVioR I don't think we can remove `NIMBUS-REASSIGN` as it used
mostly during supervisor_test to avoid rescheduling with value `false`. The
documentation clear
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150702270
@HeartSaVioR Removed rebalance calls at places where I could use simulated
time cluster.
---
If your project is set up for it, you can reply to this email and have
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150701869
@kishorvpatil
Good, I didn't think about mocking. Your additional changes look great.
Just I wish to discuss how we can do with ```nimbus.reassign```.
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150699962
@HeartSaVioR I tried that for supervisor_test.clj, somehow mocking of
submit topologies etc.methods don't make it easy. It seems straight forward to
rebalance and p
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150699176
@kishorvpatil
Seems like some places in the tests still rely on rebalance. Could we
change these to use simulated-time cluster as well when possible?
---
If you
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150697902
@kishorvpatil
Test changes look good.
@kishorvpatil @revans2
I'd like to see my previous comment
(https://github.com/apache/storm/pull/810#issuecomm
Github user revans2 commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150671588
The changes look good I am +1, and time travis took is about the same as
before.
---
If your project is set up for it, you can reply to this email and have your
reply ap
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150667453
@HeartSaVioR Thank you for review and suggestions. I have made changes so
that our tests are using simulated-time cluster to avoid `Thread/sleep`. Please
revisit the
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150051486
@kishorvpatil
rebalance seems not work cause it requires topology to be alive at the
moment.
---
If your project is set up for it, you can reply to this email a
Github user kishorvpatil commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150049582
@HeartSaVioR looking into compilation issue. Somehow it did not get
committed. Re-running the build.
---
If your project is set up for it, you can reply to this em
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150047429
I also love the concept.
Btw, in order to rely on recurring mk-assignments,
- We should remove ```nimbus.reassign``` since it should be true for new
topo
Github user HeartSaVioR commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-150045478
@kishorvpatil
You may want to check Travis build failure, seems like there's a missing
spot.
> Exception in thread "main" java.lang.IllegalArgumentExcept
Github user revans2 commented on the pull request:
https://github.com/apache/storm/pull/810#issuecomment-149996906
I love the concept, but I would like to see the tests updated so we are not
adding several mins to the time it takes to run the unit tests.
---
If your project is set up
Github user revans2 commented on a diff in the pull request:
https://github.com/apache/storm/pull/810#discussion_r42666392
--- Diff: storm-core/test/clj/backtype/storm/integration_test.clj ---
@@ -236,6 +237,7 @@
"acking-test1"
GitHub user kishorvpatil opened a pull request:
https://github.com/apache/storm/pull/810
[STORM-1121] Remove method call to avoid overhead during topology
submission time
Nimbus calls mk-assignments from SubmitTopology within lock is causing it
to wait for processing of all heartbe
21 matches
Mail list logo
