When SSPR application is left idle for around 4-5 hours I am getting the error "RequestTimeoutError: Timeout exceeded". After trying 5-6 times am able to login to the application without any issues.

DEV is working fine but the problem is with QA and PROD. The difference between DEv and QA/PROD is in DEV all servers (SSPR,IDV(user store), Oracle(store for response)) are in the app tier. But in QA / PRD SSSPR servers are in DMZ.

Username search filter used is (&(objectClass=person)(cn=%USERNAME%)(!(lockedByIn truder=true)))

Below is the snippet from the trace level log

2016-04-03T14:44:18Z, TRACE, http.PwmRequest, {5w} POST request for: /sspr/private/Login []
pwmFormID='EGTpgtfHxeWngxenYlKGxvxQ4EXeRXYF153dc92 d917iIZFl'
2016-04-03T14:44:18Z, TRACE, ldap.UserSearchEngine, {5w} username '******' does not appear to be a DN (does not start with configured ldap naming attribute 'cn') []
2016-04-03T14:44:18Z, DEBUG, ldap.UserSearchEngine, {5w} beginning user search process []
2016-04-03T14:44:18Z, DEBUG, ldap.UserSearchEngine, {5w} performing ldap search for user; searchID=5 profile=IDV base=ou=users,o=***** filter=SearchHelper: filter: (&(objectClass=person)(cn=******)(!(lockedByIntrud er=true))), scope: SUBTREE, attributes: [] []
2016-04-03T14:45:15Z, TRACE, http.PwmRequest, {5v} POST request for: /sspr/public/CommandServlet []
pwmFormID='ihobI156Ug5hAl1hjsbQkGYA6uKrWHIB153dc83 9260qdz42'
2016-04-03T14:45:15Z, TRACE, servlet.CommandServlet, {5v} received request for action idleUpdate []