Benoit Tellier created JAMES-4048:
-------------------------------------
Summary: Improve LDAP connection management
Key: JAMES-4048
URL: https://issues.apache.org/jira/browse/JAMES-4048
Project: James Server
Issue Type: Improvement
Reporter: Benoit Tellier
h3. Why?
*Post mortem notes upon some LDAP related errors*:
FYI I linked some LDAP instabilities to ASF James:
- The james errors with LDAP are connection timeout
- No default setup of OpenLDAP being done the default of 16 connections
applies (across all apps, and all servers) so total 48 connections (including
lemon, sabre, openpaas, james...)
- No idea about how to monitor LDAP connection counts both on OpenLDAP and
jamesside - I may add a metric on James side...
- The errors are linked to some traffic increase on Apache James side (couple
100 email in one minute).
- Thus it do not seem far fetched to think there's a worlf with Apache James
connection management.
- And indeed, this set up is TRICKY with our LDAP library: maxConnections is
not the maximum number of connection but the maximum number of connection
s*retained in the pool* 
Connections are created on-demand upon traffic and are unbounded (!!!) 
Meaning James can easily overload LDAP
While the inner working might not be haltered, I can effectively force a
delay before opening a new connection, wich hopefully would bring the behaviour
closer to a regular pool!
h3. How?
-> Metric collection for the pool statistics in order not to be blind!
-> Configuration of pooling parameters for the LDAP connection (max wait time,
max connection count)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
