Unexpected realtime delay error on new Mesa 7i96 install

Apologies and correction - the red "INIT" LED (CR11) becomes lit on the error. The yellow "PWR" LED is always lit! (I was not in the shop when I wrote that last email. Now logged into the forum on my Linux machine next to the board.....).

A solid red init LED means a watchdog bite, which likely means the the host
PC has decided to wander off and do something else for a fair number of
milliseconds. It might be a hardware issue (Ive seen this happen with failing
hard drives) or just some power management option that cannot be turned off

OK, sounds serious. I had the watchdog set for 50ms, and just set it for 500ms as a test. It still lights the red LED on starting LinuxCNC. It's a brand new SSD, so is this looking like I am flogging a dead horse...?

It doesn't sound good, maybe that PC is just not suited to real time

Does dmesg show any Ethernet link up/down messages?
that's another possible cause of losing communications...

can you tell us what pc this is
what Linux you are running
Did you hidde the othere processors if it has more so it runs only on one with the realtime kernel ISOCPU
very rear to get this numbers so hard
there are 50USD pc that get it down to <5k

bbsr_5a wrote: can you tell us what pc this is
what Linux you are running

He said that on the first post:
Asus EEE, if you google that you will find it is an old very small 7" laptop, at least the first series was, i had 2 of them.

@ tom what latency you got on this or did you only use them as WIN desk

bbsr_5a wrote: @ tom what latency you got on this or did you only use them as WIN desk

I had 2 of them when they got out, the first series of them. They were underwhelming so i sold them promptly, and had to install winxp on one of them for the client.
They could do over 4 hours on batteries at that time, so they had weak processors, slow memory, to much power saving, etc. They will most probably not do for Linuxcnc.

I just tried dmesg as suggested - I'm only a Linux beginner, so didn't know what to expect. But I think this may be helpful.

Before invoking LinuxCNC, it had a mix of various messages about the system. I fired up LinuxCNC, and let it go it its error state - and then retried dmesg. I got literally screenfulls of "UDP Checksum error", which sound extremely suspicious!

FWIW, I closed LinuxCNC, power cycled the 7i96, and pinged it. It worked OK, with the usual 0.3ms ping time. I tried dmesg again, but just saw mention of the link being lost and regained - due to the power cycle. So it was definitely running LinuxCNC that provoked the errors.

Any idea on what could be causing that?

Ref my PC....

They are probably what you had, Tom, but maybe in a slightly different format. They are super small form factor desktops rather than laptops, and based on an Intel Atom N270. This is a 1.8GHz single core processor of 2008 vintage. I agree that they are likely not up to the task in hand here - but they would have been brilliant if they had been, as they are extremely neat!

Funny how they are tantalisingly close though. OK, so the ping times are poor, but from what has been said the latency figures seem workable if not ideal. Even if I could have tested my motors it would have been nice, but I'm coming round to the fact that I should probably just give up on them, and move on.....