[
https://issues.apache.org/jira/browse/ACCUMULO-4510?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15837960#comment-15837960
]
ASF GitHub Bot commented on ACCUMULO-4510:
------------------------------------------
Github user keith-turner commented on a diff in the pull request:
https://github.com/apache/accumulo-testing/pull/2#discussion_r97805152
--- Diff: README.md ---
@@ -46,9 +46,88 @@ walkers in 5 containers in YARN using the Image.xml
module.
./bin/accumulo-testing rw-yarn 5 Image.xml
-This command will create an application in YARN and exit when all
containers for the test have started.
-While its running, you can view logs for each random walker using the YARN
resource manager. The YARN
-application can be killed at any time using the YARN resource manager or
command line tool.
+This command will create an application in YARN and exit when all
containers for the test have
+started. While its running, you can view logs for each random walker using
the YARN resource manager.
+The YARN application can be killed at any time using the YARN resource
manager or command line tool.
+
+## Continuous Ingest & Query
+
+The Continuous Ingest test consists of several applications that place a
continuous query and ingest
+load on Accumulo to see if it breaks.
--- End diff --
I think it would be useful to state that CI creates lots of large linked
list and the test suite can look for holes in these linked list to see if
Accumulo lost data. Could pull from [goraci]'s background section.
[goraci]: https://github.com/keith-turner/goraci
> Accumulo test benchmarks should go in a separate git repo
> ---------------------------------------------------------
>
> Key: ACCUMULO-4510
> URL: https://issues.apache.org/jira/browse/ACCUMULO-4510
> Project: Accumulo
> Issue Type: Task
> Reporter: Christopher Tubbs
> Assignee: Mike Walch
> Priority: Minor
> Time Spent: 1h
> Remaining Estimate: 0h
>
> Benchmarks should be version-independent, and/or work across versions to be
> meaningful. The tests/benchmark scripts which accompany Accumulo releases are
> also not necessarily tested or even functional. They should be shipped
> separately, if we continue to maintain them at all. I actually don't even
> think they need to be released. They can be continuously updated as needed
> for developer testing/benchmarking, without necessarily releasing (similar
> with examples, and website documentation).
> We can request a new git repo easily at https://reporeq.apache.org
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
