A slapd hangs up when many entries information were modified on RHEL5

Issue

• slapd hangs after being restarted while an operation modifying many entries is being run. The slapd's core file captured by gcore command says as follows:

  • A worker-thread is waiting for mutex.
  • A thread that locked the mutex is already gone.

• This happens when:

  • slapd exits without cleaning up the mutex, so the backend DB is in a dirty state.
  • Admin runs slaptest or slapcat.
  • service ldap start

• This issue does not manifest if slaptest or slapcat is not run.