Successfully configured with stepconf: but +Y-axis "stutters" running G-code

I managed to fix booting and even have Debian 10 with the 4.19 rtai kernel booting as the default.

The problem is that neither latency-test nor latency-histogram will run.

latency-histogram says:
insmod: ERROR: could not insert module /usr/realtime-4.19.195-rtai-amd64/modules/rtai_hal.ko: Unknown symbol in module

latency-test says:
insmod: ERROR: could not insert module /usr/realtime-4.19.195-rtai-amd64/modules/rtai_hal.ko: File exists

Running dmesg as directed:
thread: Unknown symbol hal_create_thread (err-2)
thread: Unknown symbol hal_ready (err-2)
thread: Unknown symbol hal_exit (err-2)
thread: Unknown symbol rtapi_print_msg (err-2)
thread: Unknown symbol hal_init (err-2)
 

% uname -a
Linux aquino 4.19.195-rtai-amd64 #5 SMP PREEMPT Sun Jul 11 19:13:27 BST 2021 x86_64 GNU/Linux

That's what I expected.

I wonder if I started with a bastard version of Debian 10. What is there is what was there before the motherboard crash, but because my initial efforts were directed at getting Linux Mint 20.3 RT working, I never intended to revisit it. It's possible I acted on that intention and mucked something.

My best bet is to do a clean (re)install of the recommended version for LxCNC 2.8 and do the RTAI kernel install on top of that.

Andy,
Upon your remark that the RTAI kernel and LinuxCNC ought to work in Mint as well, I thought to give that a try before scrapping the Mint install.

I remembered then that the Mint was built upon the 5.4 kernel and concluded that this eliminates it for modification with a 4.19 RTAI kernel.

Would that be right?  I'd need a 5.4 RTAI kernel to modify my Mint 20.3 install?

[Update: rtai.org released RTAI 5.3 back in May 2021 which I'll guess will not suffice.]

Mint20.3% uname -a
Linux aquino 5.4.177-rt69 #2 SMP PREEMPT_RT Sun Feb 20 21:43:22 CST 2022 x86_64 x86_64 x86_64 GNU/Linux

As I recall, and details are fading, I hunted for a late-model PREEMPT_RT kernel first and then downloaded source for that version of the OS before compiling.  Some OS versions were not candidates because no corresponding RT kernel was found.

[Update: rtai.org released RTAI 5.3 back in May 2021.  I presume it won't work for a 5.4 kernel.]

I am currently downloading linuxcnc-2.8.2-buster.iso from:
www.linuxcnc.org/iso/

which I understand will have a PREEMPT_RT kernel (4.19 I presume) so I will install the 4.19 RTAI kernel and LinuxCNC afterwards.

I downloaded
linuxcnc-2.8.2-buster.iso
and burned a DVD and booted i.

I did a graphic install overtop the existing Deb10 partitions (I have a /boot and a /root) without altering size, although I did direct them to be reformatted. 

There was a bit of panic on my part when I saw DHCPv6 come up, because I don't run v6, but it seemed to get by that and continued to the end. 

The only real difference was that I asked GRUB to be put on the second disk, /dev/sdb, where these new Deb10 partitions are. (There are three hard drives, and at one point I had the bootloader identically on all three.) 

When the install finished, I went to boot the hdd and it failed, saying it couldn't find a partition.  I hadn't simply accepted the default boot option, but looked at the proposed Debian 10 menu item with the 'e' option--and noticed that it was going to boot an "rtai" kernel! That seems like a holdover from the previous install. 

I had definitely instructed the installer to reformat both the /boot and /root partitions and even to zero-out the /boot partition. (None of the Deb 10 GRUB entries on any of the disks' boot menus mentioned the rt image: only rtai.) 

I was indeed booting from the /dev/sdb because I accessed a Boot Menu during booting (in fact I've tried booting from all three) so I just decided to "Start Again." 

This time, the DVD drive boots but after I select Graphic Install (or Install), it hangs at the DHCPv6 step: 30 min and counting at this point. It shows an empty blue screen with an empty white bar below. I doubt it will come back unless a DHCPv6 server spontaneously appears.  

I'm going to try to boot into a Live session and give it a fixed IP address, however that's done, and then launch the install again from there.

Can you tell me the pathname to the installer?

It was actually easier than configuring a static IP address; I just unplugged the network cable, which caused the installer to move on to trying DHCP, at which point I quickly plugged it back in, and a dynamic address was fetched.

Before I got to this point, I remembered the console terminals and dropped down to one (Ctrl-Alt-F2) to run a 'ps' to see if I could detect the installer.

Sure enough it's running as '/bin/sh /sbin/debian-installer' iirc.

At the moment, I've gotten past the manual partitioner where again I have asked that sdb7 (/root) and sdb8 (/boot) be reformatted.  It's done that and is copying files.

When it's done, I'll let it put the bootloader on the first hdd's MBR--OR WHEREVER IT WANTS TO. I'm easy, and we'll see if it'll boot of the hard drive this time.  Before I let it reboot, I'll drop down to a console again and try to see if /dev/sdb8 actually has bootable -rt- image files.

I repeated the install with the steps and decisions I had taken the first time.

It has failed again. 

I've attached a pic of the boot entry in grub for the Debian 10 os on drive /dev/sdb7 (hd1; msdos) and for the subsequent "partition not found" error message when that entry is selected.

Note the peculiarity in the grub entry referring to an 'rtai' image.  Note also in that pic the long hex number which is referenced in the second.  Once again, I called for both the /boot and the /root partitions (sdb7/sdb8) to be formatted. 

The reference to the long-hex in the second screen leads me to believe I should have somehow enabled the option signified by a '?' on the first screen.

Bizarre.