Systemd sd-pam high memory during tally

[hansmiddelhoek]

On one of our servers LFD is reporting high memory usage for /lib/systemd/systemd (sd-pam) during tally:

Time:         Fri Jul 15 00:16:07 2022 +0200
Account:      USER_A
Resource:     Virtual Memory Size
Exceeded:     1132 > 512 (MB)
Executable:   /lib/systemd/systemd
Command Line: (sd-pam)

2022:07:15-00:15:54: Tally User USER_A Begin
2022:07:15-00:16:55: Tally User USER_A Complete

Details of USER_A:
* 0 databases
* 256MB disk usage
* only a few MB of bandwidth

This happens for 5 different users on this server. I don't see this on other servers. This server has only 25 users in total.

OS: Debian 9.13
Version: DirectAdmin v.1.641
package: directadmin_d1f1ed9f3fda0faa809e53b522e130d5083ccd39_debian9_amd64.tar.gz
systemd 232
Webserver: nginx_apache

This happens every day since I updated to DA 1.641, probably this would have happened in earlier versions but this server was running an older DA version for a long time and is updated directly to 1.641. Why does sd-pam uses so much memory during the tally process? Can this be solved? Especially because this is happening on such a small account.

I can't find anything as a possible cause for this. The only thing is that all of these users have multiple domains where some of them aren't used and have (webserver) logs of 0 bytes. Maybe something in the tally code that flips on empty logs??

Thanks!

 

[factor]

Have you tried a

apt-get update && apt-get upgrade

then do

cd /usr/local/directadmin/custombuild
./build set_fastest
./build clean
./build update
./build all
./build rewrite_confs

sd-pam

I mean it is systemd... there was a bug report on Jessie for this.

during the tally process

You could work through this.

Task queue processor | Directadmin Docs

I would just upgrade to 10 > 11. Being a bit more current on the OS side might help.