-
Type:
Bug
-
Resolution: Timed out
-
Priority:
Low
-
None
-
Affects Version/s: None
-
Component/s: User Management, Crowd & LDAP integration
-
Severity 3 - Minor
LDAP pooling is not working properly in Bamboo. We have the following set in our atlassian-user.xml file.
<securityProtocol>ssl</securityProtocol> <poolingOn>true</poolingOn> <maxSize>100</maxSize> <initSize>1</initSize> <prefSize>5</prefSize> <debugLevel>none</debugLevel> <timeout>300000</timeout> <initialContextFactory>com.sun.jndi.ldap.LdapCtxFactory</initialContextFactory> <batchSize>100</batchSize> <timeToLive>0</timeToLive> <userSearchAllDepths>true</userSearchAllDepths> <groupSearchAllDepths>true</groupSearchAllDepths>
When we go to the Groups page in the Bamboo UI about 20 connections to our ldap server is made and remain established connections. After waiting 15 seconds once the first page of groups to appear, I click on the page 2. You would think the 20 connections in the LDAP pool would be reused since they are already open and established but that is not the case. Instead it opens another 20 so now there is 40. This keeps occurring until the pool reaches 100. Then an error in the log is thrown and I can no longer access the groups page in bamboo until the timeout occurs (5 minutes).
2016-04-21 16:52:41,763 ERROR [http-nio-8085-exec-10] [LDAPMembershipToUsernamePager] Error converting search result: cn=some-group: null:null: {memberuid=memberUid: uid=xxxxxxxxxxxxxxxxxx} into list of members as usernames. com.atlassian.user.impl.ldap.repository.LdapConnectionFailedException: javax.naming.CommunicationException: Timeout exceeded while waiting for a connection: 30000ms
There are two issues here. The first issue is the error. Once the maxSize is reached on the pool, should it even be trying to connect to the LDAP server? Should a friendly message occur stating the pool max has been reached?
The second issue is that pooled LDAP connections are not being reused. This defeats the whole purpose of using the pool.
You can see that amount of connections growths when user paginates over groups (~100 groups)
netstat -na | grep -c 636
Issue doesn't exists for browsing users (~500 users).
- relates to
-
BDEV-11866 Loading...