[
https://issues.apache.org/jira/browse/SENTRY-634?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14302689#comment-14302689
]
Anne Yu commented on SENTRY-634:
--------------------------------
Put more context into this JIRA. We found one performance issue when more than
500+ clients are trying to connect to sentry at the same time. What happened is
Sentry service was down, other services on the same hosts were also brought
down. So scaling tests are helpful in this case:
1) If the #connections is configured to a real large number, Thank [~colin_mjj]
for pointing it out, how much scale Sentry service can go up, based on the host
resources. What other factors affect the performances;
2) When boundary conditions are reached, what the behavior of the Sentry
service is like? Hang, start to reject connections or restart itself etc.
> Scability Test: create > 500 connections, even more, to see if sentry service
> can handle client requests
> --------------------------------------------------------------------------------------------------------
>
> Key: SENTRY-634
> URL: https://issues.apache.org/jira/browse/SENTRY-634
> Project: Sentry
> Issue Type: Test
> Affects Versions: 1.4.0
> Reporter: Anne Yu
> Fix For: 1.4.0
>
>
> 539 2015-01-28 08:10:00,031 WARN
> sentry.org.apache.thrift.server.TThreadPoolServer: ExecutorService rejected
> client 1009 times(s) 27540 java.util.concurrent.RejectedExecutionException:
> Task sentry.org.apache.thrift.server.TThreadPoolServer$WorkerProcess@30e8e6ab
> rejected from java.util.concurrent.ThreadPoolExecutor@73451032[Running, pool
> size = 500, active threads = 500, queued tasks = 0, completed tasks = 365]
> 27541 at
> java.util.concurrent.ThreadPoolExecutor$AbortPolicy.rejectedExecution(ThreadPoolExecutor.java:2047)
> 27542 at
> java.util.concurrent.ThreadPoolExecutor.reject(ThreadPoolExecutor.java:83)
> 27543 at
> java.util.concurrent.ThreadPoolExecutor.execute(ThreadPoolExecutor.java:1369)
> 27544 at
> sentry.org.apache.thrift.server.TThreadPoolServer.serve(TThreadPoolServer.java:136)
> 27545 at
> org.apache.sentry.service.thrift.SentryService.runServer(SentryService.java:232)
> 27546 at
> org.apache.sentry.service.thrift.SentryService.access$000(SentryService.java:72)
> 27547 at
> org.apache.sentry.service.thrift.SentryService$2.run(SentryService.java:159)
> 27548 at
> org.apache.sentry.service.thrift.SentryService$2.run(SentryService.java:156)
> 27549 at java.security.AccessController.doPrivileged(Native Method)
> 27550 at javax.security.auth.Subject.doAs(Subject.java:422)
> 27551 at
> org.apache.sentry.service.thrift.SentryService.call(SentryService.java:156)
> 27552 at
> org.apache.sentry.service.thrift.SentryService.call(SentryService.java:72)
> 27553 at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> 27554 at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> 27555 at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> 27556 at java.lang.Thread.run(Thread.java:745)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
