[Documentation enhancement] Server hangs due to rsyslog running out of queue space

[rhessing]

I've noticed that when the remote syslog server is not available rsyslog will 'hang on' to the logs until it is back online. However, if the queue runs out of queue space then the inputs begin to block, which causes the log socket to block and finally the apps (including the kernel) that write to the log socket will hang causing a system wide hang.

According to other issues in regards to this topic this is the default rsyslog behavior unless rsyslog is configured in such a way that it knows when to discard log messages or write it to local disk.

Could we please add this behavior to the documentation? I only found a short mention here:
https://www.rsyslog.com/doc/master/rainerscript/queue_parameters.html?highlight=hang#queue-timeoutenqueue

But I believe this requires a more prominent position within the documentation such as here:
https://www.rsyslog.com/doc/master/troubleshooting/index.html#typical-problems

[rgerhards]

However, if the queue runs out of queue space then the inputs begin to block, which causes the log socket to block

It's not actually blocking, but the main queue timeout is 2sec (per msg). So it drains veeery slowly.

But I believe this requires a more prominent position within the documentation such as here:
https://www.rsyslog.com/doc/master/troubleshooting/index.html#typical-problems

ack - are you up for a PR?

[rgerhards]

doc source is here, just in case: https://github.com/rsyslog/rsyslog-doc

[rhessing]

Sure, I'll create a PR