ORBI PRO SRK60 - Syslog

terminaljunkie For what is offered as a Pro device for SOHO and SMB I agree - for audit reasons - that the router application log messages must be sent to the syslog mainly because these are not going to a flash, but to a volatile memory only.

This is part of Netgear's and many other router makers decades old design, typically paired with the ability to email - what a useless idea - to share the logs if the buffer is full or on a defined schedule. And long before Netgear started to make their own. When I remember right I admit I had promoted this idea to a major modem maker around at the time when POTS and ISDN modems started to make their first steps to become dial-up routers in 1995 or early 1996. There was simply no better commonly available better choice back then. 

I've suggested Netgear a loooong time ago on how to enhance the event handling system a long time ago in a flexible way where the user could define what kind of data to be collected and to what kind of destination it should be sent to ... no luck, no progress. And yes, this includes the control for allowing to feed the application events and select Kernel output eg. to the syslog.

Much more useful information about the ongoing activities from these embedded Linux systems could be taken from two sources: 

• syslog ... permitting all processes and features do feed to it (nooo this is in no way "meaningless gibberish " as you state!), and
• the Kernel output from the console, which does go to a ring buffer (volatile, too)

Evaluating the console/Kernel output (where the most useful information on a process crash cause would show up!) is kind of clueless, in the case of a crash it's very unlikely that the process supposed to forward eg. useful and informative data like OOM and process crash conditions are able to continue, so it would be lost anyway. Except where dedicated physical serial console management systems are still in place - what would be ways beyond of what the average SOHO or SMB does operate in 2021 - something given up in many mid and large scale data centers for cost reasons. So we're back there on the syslog for U**x-lik systems and the Windows audit and event log (shamelessly borrowed from OpenVMS). 

Back to these small embedded routers: The mass of information does make it difficult to impossible to log into a non-volatile memory ... for cost reasons, and for reliability reasons. The small flash section won't make it long. So it's hard to keep useful post-mortem information...