Alexey Serbin has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/9526 )

Change subject: A new Jepsen checker for READ_YOUR_WRITES scan mode
......................................................................


Patch Set 9:

(7 comments)

http://gerrit.cloudera.org:8080/#/c/9526/9//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/9526/9//COMMIT_MSG@10
PS9, Line 10: unique value
unique values?

I.e., as I understand from the description, every client writes multiple 
values, and those values are unique across all active clients, right?


http://gerrit.cloudera.org:8080/#/c/9526/9//COMMIT_MSG@12
PS9, Line 12: ,
:


http://gerrit.cloudera.org:8080/#/c/9526/9//COMMIT_MSG@12
PS9, Line 12: 'count', 'add'
It would be nice to describe what 'count' and 'add' mean, in essence.


http://gerrit.cloudera.org:8080/#/c/9526/9//COMMIT_MSG@13
PS9, Line 13: than or equal to the
            : count of previous writes by the same client, and the row count 
never go
            : down from the previous reads
Is there any other mode in which that constraint would be broken?

I mean, by the way how the clients behave, I cannot imagine there would be some 
different result output.  Do I miss anything?


http://gerrit.cloudera.org:8080/#/c/9526/9/java/kudu-jepsen/src/main/clojure/jepsen/kudu/sets.clj
File java/kudu-jepsen/src/main/clojure/jepsen/kudu/sets.clj:

http://gerrit.cloudera.org:8080/#/c/9526/9/java/kudu-jepsen/src/main/clojure/jepsen/kudu/sets.clj@38
PS9, Line 38: ;;   [{:type :invoke, :f :add, :value 1, :process 0}
            : ;;    {:type :ok, :f :add, :value 1, :process 0}
            : ;;    {:type :invoke, :f :add, :value 2, :process 0}
            : ;;    {:type :ok, :f :add, :value 2, :process 0}
            : ;;    {:type :invoke, :f :count, :value nil, :process 0}
            : ;;    {:type :ok, :f :count, :value 2, :process 0}
            : ;;    {:type :invoke, :f :count, :value nil, :process 1}
            : ;;    {:type :ok, :f :count, :value 2, :process 1}
            : ;;    {:type :invoke, :f :add, :value 13, :process 1}
            : ;:    {:type :ok, :f :add, :value 13, :process 1}
            : ;;    {:type :invoke, :f :add, :value 22, :process 0}
            : ;;    {:type :ok, :f :add, :value 22, :process 0}
            : ;;    {:type :invoke, :f :count, :value nil, :process 0}
            : ;;    {:type :ok, :f :count, :value 2, :process 0}
            : ;;    {:type :invoke, :f :add, :value 250, :process 1}
            : ;;    {:type :ok, :f :add, :value 250, :process 1}]
            : ;;
> why? what's the anomaly that the checker is flagging?
Yep, I agree with David: it would be nice to describe the problem in the output.

Is that something related to

;;    {:type :invoke, :f :add, :value 22, :process 0}
;;    {:type :ok, :f :add, :value 22, :process 0}
;;    {:type :invoke, :f :count, :value nil, :process 0}
;;    {:type :ok, :f :count, :value 2, :process 0}

?

I.e., the expected count output should be 3 (given the history of the 
operations above), but the actual output was 2.  Or it's something else?


http://gerrit.cloudera.org:8080/#/c/9526/9/java/kudu-jepsen/src/main/clojure/jepsen/kudu/sets.clj@198
PS9, Line 198: READ_YOUR_WRITES
Is there a mode that, if specified, will make this test fail at all?


http://gerrit.cloudera.org:8080/#/c/9526/9/java/kudu-jepsen/src/main/clojure/jepsen/kudu/sets.clj@203
PS9, Line 203: value
values



--
To view, visit http://gerrit.cloudera.org:8080/9526
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I92d5c0e3b91af58576eb6cd408d922ec7c0fef6c
Gerrit-Change-Number: 9526
Gerrit-PatchSet: 9
Gerrit-Owner: Hao Hao <[email protected]>
Gerrit-Reviewer: Alexey Serbin <[email protected]>
Gerrit-Reviewer: David Ribeiro Alves <[email protected]>
Gerrit-Reviewer: Hao Hao <[email protected]>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Comment-Date: Tue, 08 May 2018 01:33:58 +0000
Gerrit-HasComments: Yes

Reply via email to