[
https://issues.apache.org/jira/browse/DIRSERVER-2369?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17520419#comment-17520419
]
Alexander Schwartz commented on DIRSERVER-2369:
-----------------------------------------------
Thanks [~elecharny] for commenting on this. Looking at this problem again, I
see that I was led astray by an error message in our code.
I was analyzing a (probable) race condition where a cleanup of some LDAP roles
in tests fail in. After changing this to keepalive connections, I first thought
it wouldn't happen any more, after further tests I see just happens less often
(once in about 150 times)
I still don't know why this fails. The log shows (once in about 150 times) the
message
{{Caused by: javax.naming.NamingException: [LDAP: error code 1 -
OPERATIONS_ERROR: failed for MessageType : SEARCH_REQUEST}}
{{Message ID : 2}}
{{SearchRequest}}
{{baseDn : 'cn=non-returned-role-4,ou=RealmRoles,dc=keycloak,dc=org'}}
{{filter : '(objectClass=*)'}}
{{scope : single level}}
{{typesOnly : false}}
{{Size Limit : no limit}}
{{Time Limit : no limit}}
{{Deref Aliases : deref Always}}
{{attributes :}}
{{org.apache.directory.api.ldap.model.message.SearchRequestImpl@cfe6acb0
ManageDsaITImpl Control}}
{{Type OID : '2.16.840.1.113730.3.4.2'}}
{{Criticality : 'false'}}
{{'}}
Then the next test complains that the role exists.
As there were a log of the other "UNBIND_REQUEST" messages above, I assumed
they would be the cause.
It is probably correct to close this issue, although I still need to figure out
what the cause is.
> ignoring message received from null session
> -------------------------------------------
>
> Key: DIRSERVER-2369
> URL: https://issues.apache.org/jira/browse/DIRSERVER-2369
> Project: Directory ApacheDS
> Issue Type: Bug
> Affects Versions: 2.0.0.AM26
> Reporter: Alexander Schwartz
> Priority: Major
>
> In a test setup we regularly saw the warning that an unbind request wasn't
> processed (see log snippet below). This lead to errors later in our test, as
> it assumed that the unbind would have been successful.
> Looking at the comment in the code of LdapRequestHandler, it reads: "in some
> cases the session is becoming null though the client is sending the
> UnbindRequest before closing".
> Our workaround attempt is now to enable Java's LDAP connection pooling so
> that we don't close the LDAP session right after sending the unbind request.
> Ideally we would like to see the unbind request being processes regardless of
> the closed session, as the caller doesn't really have means to determine if
> the asynchronous processing has completed.
> It would be great if ApacheDS would handle this case.
>
> The workaround for Java LDAP connection pooling:
> * com.sun.jndi.ldap.connect.pool set to true to enable connection pooling
> * com.sun.jndi.ldap.connect.pool.maxsize set to the same value as
> com.sun.jndi.ldap.connect.pool.prefsize to avoid short-lived connections
>
> The log:
> {{11:05:56,830 INFO
> [org.apache.directory.server.ldap.handlers.LdapRequestHandler]
> (pool-26-thread-1) ignoring the message MessageType : UNBIND_REQUEST}}
> {{Message ID : 3}}
> {{UnBind
> Requestorg.apache.directory.api.ldap.model.message.UnbindRequestImpl@cffb7dca
> ManageDsaITImpl Control}}
> {{Type OID : '2.16.840.1.113730.3.4.2'}}
> {{Criticality : 'false'}}
> {{'}}
> {{received from null session}}
>
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
