Description of problem:

I have some issues to troubleshoot this issue further.

In some conditions of strong load, a same query is skipping filter evaluation completely. Here it's the extract in the same connection:


[26/Oct/2015:17:42:43 -0700] conn=841921 op=10 SRCH base="ou=Email Aliases,ou=ITS,o=Caltech,c=US" scope=1 filter="(caltechITSEmailAliasAddress=zleite)" attrs="caltechITSEmailAliasDestination"
[26/Oct/2015:17:42:54 -0700] conn=841921 op=10 RESULT err=0 tag=101 nentries=24804 etime=16 notes=U
[26/Oct/2015:17:43:15 -0700] conn=841921 op=11 SRCH base="ou=Accounts,ou=ITS,o=Caltech,c=US" scope=1 filter="(uid=zleite)" attrs="uid"
[26/Oct/2015:17:43:15 -0700] conn=841921 op=11 RESULT err=0 tag=101 nentries=0 etime=4
[26/Oct/2015:17:43:29 -0700] conn=841921 op=12 SRCH base="ou=Accounts,ou=ITS,o=Caltech,c=US" scope=1 filter="(caltechITSEmailAliasAddress=zleite)" attrs="uid"
[26/Oct/2015:17:43:32 -0700] conn=841921 op=12 RESULT err=0 tag=101 nentries=0 etime=9 notes=U
[26/Oct/2015:17:43:42 -0700] conn=841921 op=13 SRCH base="ou=Email Aliases,ou=ITS,o=Caltech,c=US" scope=1 filter="(caltechITSEmailAliasAddress=zleite)" attrs="caltechITSEmailAliasDestination"
[26/Oct/2015:17:43:50 -0700] conn=841921 op=13 RESULT err=0 tag=101 nentries=0 etime=12 notes=U

Note operation op=10 and op=13 is the same. One of them returning 0 entries and the other one returning nentries=24804 as if the filter is not used.

These queries are doing full scan queries. Customer is not adding an index but in any case, this is a bug that has to be solved independently of performance. 

We cannot add "filter access" error log level because customer is in production.

Please, anyway I could troubleshoot this independently of filter errorlog level, it's welcome.

I am attaching the error logs to this case.


Version-Release number of selected component (if applicable): 

389-ds-base-1.3.3.1-20.el7_1.x86_64