This bug was initially created as a copy of Bug #1925559 in order to achieve a backport to  Red Hat Enterprise Linux 8 as well.

Description of problem:

A customer had an issue for which the sssd queries were taking 2.5 seconds to be fulfilled, while the usual time in their environment was 0.1 seconds.
In the sssd logs they were not able to spot any problem on the LDAP backend.

The customer would like to introduce a timer for any LDAP query, which should be written in the debug logs (correct debug level to be identified).
They would like also to have these queries logged into the sssd with a warning level (lower than debug) once the time taken is really high (eg 5 seconds).
The time for which the logs should be written in warning level in place of debug level should be configurable.

This would help the debug of the LDAP backend during incidents (eg network or LDAP backend overload/service disruption) and to identify deviation from the regular query time from a given server.
bonus: this metric could be useful also for the PCP project.


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


How reproducible:

Introduce a network latency or LDAP backend latency between sssd and LDAP backends.

Steps to Reproduce:

1. introduce a network latency or LDAP backend latency between sssd and LDAP backends
2. increment the logs to debug=9
3. check the sssd logs

Actual results:
No information over the time taken for a given query

Expected results:
Query time is written in the debug logs,
Long time queries written in the regular logs.