[CONJ-256] Intermittent connection failure with MaxScale 1.2/1.3 using MariaDB/J 1.3 Created: 2016-02-18  Updated: 2016-02-19  Resolved: 2016-02-19

Status: Closed
Project: MariaDB Connector/J
Component/s: Other
Affects Version/s: None
Fix Version/s: 1.3.6

Type: Bug Priority: Major
Reporter: Guillaume Lefranc Assignee: Diego Dupin
Resolution: Not a Bug Votes: 0
Labels: None

Issue Links:
Blocks
blocks MXS-585 Intermittent connection failure with ... Closed
Problem/Incident
is caused by MXS-585 Intermittent connection failure with ... Closed

 Description   

Environment:
3 nodes of MariaDB Galera Cluster
1.2.1 or 1.3 MaxScale default configuration with galera monitor, RW split and readconnroute

Test case with Simple python2 script (install jaydebeapi lib with pip first)

import jaydebeapi
 
 
 
for i in range(1,10000):
 
    conn = jaydebeapi.connect('org.mariadb.jdbc.Driver', ['jdbc:mariadb://192.168.99.100:4008/option_service', 'root', 'admin'],'../mariadb-java-client-1.3.4.jar',)
 
    curs = conn.cursor()
 
    curs.execute('select ' + str(i) + ';')
 
    data = curs.fetchall()
 
    for val in data:
 
        print val[0].value
 
    conn.close()

After some iterations:

1
 
...
 
430
 
Traceback (most recent call last):
 
  File "jdbc.py", line 4, in <module>
 
    conn = jaydebeapi.connect('org.mariadb.jdbc.Driver', ['jdbc:mariadb://192.168.50.104:4008/test', 'maxuser', 'maxpwd'],'mariadb-java-client-1.3.4.jar',)
 
  File "/usr/lib/python2.7/site-packages/jaydebeapi/__init__.py", line 359, in connect
 
    jconn = _jdbc_connect(jclassname, jars, libs, *driver_args)
 
  File "/usr/lib/python2.7/site-packages/jaydebeapi/__init__.py", line 183, in _jdbc_connect_jpype
 
    return jpype.java.sql.DriverManager.getConnection(*driver_args)
 
jpype._jexception.SQLInvalidAuthorizationSpecExceptionPyRaisable: java.sql.SQLInvalidAuthorizationSpecException: Could not connect: Access denied for user ''@'192.168.50.1' (using password: NO)
 
 

I posted an issue in MaxScale since this regards both connector and MaxScale
Bug is not reproducible with Connector/J 1.2 and does not happen when connecting directly to the server

 Comments   
Comment by markus makela [ 2016-02-18 ]

I can confirm that this was a MaxScale problem.

Versions from at least 1.3.3 seem to do something different as MaxScale receives the packet header in the first network packet and the payload in the next.

Comment by Diego Dupin [ 2016-02-19 ]

Various tests were performed to reproduce this problem without errors.
Since markus confirm that this was a maxscale problem, i close this issue.

Generated at Thu Feb 08 03:14:20 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.