Monitoring and settings of transfers using the Linux AF_UNIX?

Question

I'm trying to setup a robust AF_UNIX data transfer.

So I make this test (on a Ubuntu 20.04): I generate logs :

for a in  `seq 1 1000000`; do echo $a | logger; echo $a  ; done

I check my log file :

tail -f /var/log/syslog | less

I can see my logs logged by rsyslog in this file.

I stop rsyslog :

systemctl stop syslog.socket

No logs arrived anymore in /var/log/syslog as expected.

But when I restart rsyslog :

systemctl start syslog.socket

I can see that the logs arrvie again but most of the one created during the rsyslog outage are missing.

I tired to increase the max_dgram_qlen

echo 5000000 > /proc/sys/net/unix/max_dgram_qlen

but it doesn't change anything.

I read on this post What is the max size of AF_UNIX datagram message in Linux? that I can monitor the socket using:

ss -ax | grep log

but the counters looked to be at zero all the time whether rsyslog is started or not.

What I trying to do :

• increase the buffer to be able to have at least 500MB or 500,000 messages in queue
• be able to monitor the size of the queue : size or queue length
• if possible logging in dmesg when messages are dropped.