starlingx/config-files
Code Issues Proposed changes

Limit size of slapd logs to critical events

Browse Source 
It has been observed that in large DC systems with thousands
of subclouds, slapd logs grow very quickly, from 2.2MB to 6.4MB
in 10 seconds. The logs also rotate once in 10min ending up keeping
just a few hours of historical logs.

This commit is changing the loglevel for slapd service to the value
"none" from "stats".
The loglevel setting of "none" does not mean that there is no logging.
Logs will contain only important events, such as start/stop ldap
server, critical errors and important messages related to the server
functionality.
The current loglevel "stats" logs all connections, operations, and
results, and while is a good starting point for diagnosing issues,
it can create a lot of messages on a busy server.
Log levels "stats" together with other verbose log levels can be set
in a lab environment when debugging issues.

Test Plan:
PASS: Verify the stx build installs correctly in an AIO-SX system
configuration and a DC system with 1K subclouds.
PASS: Verify that in a DC system with 1k subclouds the slapd logs
rotate much slower. In a system with moderate activity will rotate
about every 12 hours.
PASS: Verify that ldap commands, like ldapsearch, ldapfinger work
as expected.

Closes-Bug: 2086626

Signed-off-by: Carmen Rata <carmen.rata@windriver.com>
Change-Id: I4b6e3b14789402f10d32d341e05403894a32cc36
This commit is contained in:
Carmen Rata 2024-11-04 19:15:15 +00:00
parent 0644d3f9fa
commit 8fc354d30a
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
openldap-config/source-debian/slapd.conf
View File

@ -100,7 +100,7 @@ access to *
by self write
by * read
loglevel stats
loglevel none
overlay ppolicy
ppolicy_default "cn=default,ou=policies,dc=cgcs,dc=local"