[
https://issues.apache.org/jira/browse/IMPALA-8159?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16762696#comment-16762696
]
Jinjie Zhang edited comment on IMPALA-8159 at 2/7/19 2:02 PM:
--------------------------------------------------------------
This only happen when enable sasl authentication.
I have known the invalid sasl status reason, which is caused by a misconfigured
tableau client. When I run the test code, there is a tableau connection to
access impala service which is periodically refreshed by tableau server. The
client is configured to use http authentication protocol to access 21050 port.
Once the refresh happen, impalad thrift server will report invalid sasl status
and it will hang all thrift client.
But I don't know why one client is misconfigured, it will cause all clients
hang for serval minutes, which may be a bad design. I check the impalad
TAcceptQueueServer source code, which seems to use one thread to deal with
authentication connection.
was (Author: jinjie zhang):
This only happen when authenticating against LDAP.
I have known the invalid sasl status reason, which is caused by a misconfigured
tableau client. When I run the test code, there is a tableau connection to
access impala service which is periodically refreshed by tableau server. The
client is configured to use http authentication protocol to access 21050 port.
Once the refresh happen, impalad thrift server will report invalid sasl status
and it will hang all thrift client.
But I don't know why one client is misconfigured, it will cause all clients
hang for serval minutes, which may be a bad design. I check the impalad
TAcceptQueueServer source code, which seems to use one thread to deal with
authentication connection.
> TAcceptQueueServer: Caught TException: invalid sasl status
> ----------------------------------------------------------
>
> Key: IMPALA-8159
> URL: https://issues.apache.org/jira/browse/IMPALA-8159
> Project: IMPALA
> Issue Type: Bug
> Components: Backend
> Affects Versions: Impala 2.11.0
> Environment: python: 3.6.6
> impyla: 0.14.1
> thrift-sasl: https://github.com/yimian/thrift_sasl.git
> impala: 2.11.0 from cdh5.14.2-1.cdh5.14.2.p0.3
> os: CentOS Linux release 7.2.1511 (Core)
> Reporter: Jinjie Zhang
> Priority: Critical
>
> I enable ldap user authentication for impalad. I run the following code to
> test impalad stability, and I find when I run ten thousand times, the
> following error will occur two times in impalad.INFO log file.
> test code:
>
> {code:java}
> import impala.dbapi
> test_conf =
> { 'host': '172.16.24.xx', 'port': 21050, 'user': 'xxxx', 'password': 'xxxxx' }
> def test_impala(conf):
> conn = impala.dbapi.connect(**conf)
> conn.close()
>
> N = 10000
> for i in range(N):
> test_impala(test_conf){code}
>
> error info:
> {code:java}
> I0204 15:41:40.600821 168276 thrift-util.cc:123] TAcceptQueueServer: Caught
> TException: invalid sasl status
> I0204 15:46:40.600451 168276 thrift-util.cc:123] TAcceptQueueServer: Caught
> TException: EAGAIN (timed out)
> I0204 15:46:40.609858 168276 authentication.cc:268] Trying simple LDAP bind
> for: uid=xxxx,dc=xxxx
> I0204 15:46:40.729262 168276 authentication.cc:280] LDAP bind successful
> I0204 15:46:40.729287 168276 authentication.cc:478] Successfully
> authenticated client user "xxxx"{code}
> When this error occurs, the connect request will hang for serveral minutes,
> and all subsequent connect request including request from hue will hang too.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
