On 7/26/2022 1:56 AM, Wilkinson, Hugo (IT Dept) wrote:
If you have the ability to do so and your kernel is v5+, out-of-hours experiment with disabling system swap entirely (vm.swapiness=0 and 'swapoff -a' ) and then simulate a run of requests you'd expect to encounter the page errors with and see if it stops happening.
We use to run our LDAP servers under Gentoo with a 5.4 kernel at the time, but currently are using Rocky Linux 8 which comes with a modified 4.18 kernel.
I did actually run a production system with no swap for a while, it still occasionally had slow responses for queries requesting the memberOf attribute.
Part of the problem is I don't expect to ever see these slow response times 8-/. As of yet I have not figured out what is going on when they are slow, I definitely don't have a reproducible query test case to make it happen <sigh>.
We were running 2.4 under rocky for a while before upgrading to 2.5 and never saw this problem, so I'm pretty sure it's the result of some change in the dynlist implementation.