[IMPLEMENTED] MXdiags when run as server - log rotate

[mcrossley]

The connection is not timed out. It will run as long as the network is up, or your browser window doesn't "sleep"

[rogerthn]

OK, I think a 20MB log should zip down to around 1MB if my calcs are correct.

I'll start with 20MB and see how that goes!

Successful log rotate

Code: Select all

2020-08-25 18:40:00.443 Rotating log file, new log file will be: 20200825-184000.txt

Size 20972354 bytes.

[mcrossley]

[HansR]

Works fine. One additional remark: maybe its useful to append to the existing log file at the start of CMX i.s.o. creating a new one? (in case of backup, testing or whatever would restart CMX more often than expected)

[radilly]

Random thought I had a few days back about the service was outputting data to a telnet port.

Don't know the implications; it just came to mind while opening a terminal connection to another device.

Phil.

FYI - Syslog is a fairly standard way to externalize logging. I tinkered with it for my gateway-router, but decided there wasn't much benefit for my usage.
https://en.wikipedia.org/wiki/Syslog

[radilly]

FYI - I've been tinkering with logrotate on a Pi for some services I've written. It's a pretty widely used facility in Linux administration. I looked quickly but didn't spot a Windows port that was being actively maintained.

My logs, and I think the typical logrotate model are logs that get appended to across restarts. The rotating is independent of the service starts and stops.

Reference: https://linux.die.net/man/8/logrotate

I will say that when I wrote my own rotate function I would

• rename the log
• tail the renamed file to start the (new) log, like 100 lines

That creates some overlap that has been useful when the event would otherwise span 2 logs.


I couldn't spot how to enable this feature. Is it always on? Looks like my log runs roughly 100 KB / day when there are no issues, and I seem to be restarting often enough not to hit the limit.

Bob

[radilly]

Oh, another Unixy model familiar to systemd users is journalctl. My services leverage the journal for error conditions, and keep their own detailed logs for info and debug messages as well. I use journalctl first, and dig into the full log if I want more context.

Ref: https://manpages.debian.org/stretch/sys ... .1.en.html