On Saturday 01 December 2018 10:02:29 rhkra...@gmail.com wrote: > On Friday, November 30, 2018 07:26:33 PM Gene Heskett wrote: > > On Friday 30 November 2018 13:58:52 Michael Stone wrote: > > > On Fri, Nov 30, 2018 at 05:23:09PM +0000, Michael Thompson wrote: > > > >Because if your root partition fails, you can reinstall and all > > > > your > > > > > > > > files are safe on their own partition... > > > > > > ...leaving open the question of how likely that scenario is. > > > > Not bloody likely Michael, if the disk is toast, so are all its > > partitions as a general rule. If you are going to put you /home on a > > separate partition, put it on a different disk. > > > > Unfortunately that has NOT been acceptable to the installer for most > > of a decade now. > > Hmm, it hasn't been 10 years since I installed Jessie, yet I have my > top level directories (e.g., /abc, which hold my data directories > (e.g., /abc/Documents)) on a separate disk, and I'm rather certain I > did that with the installer -- maybe I used a different version of the > installer (or maybe I'm mis-remembering -- maybe I created those > partitions (on a separate disk) after the installation. > > But, I've thought about it for a few moments, and I'm more certain I > did that with the installer...
Maybe it has a mode that allows it, but when I last made an install useing a distro installer, it wasn't having any of that. Would not proceed past the disk partitioner point. So I got in the car and drove the 25 miles to my nearest Staples and bought a disk big enough and then installed it on the new disk, then put the old ones back in and used mc to copy my data to the new disk. I think that install was ubuntu hardy heron. Every other install since on x86 hardware has been from a 32 bit i386 iso compiled by the linuxcnc folks which is currently still wheezy based, and 32 bit because the 64 bit kernels IRQ latency is horrible due to its much larger stack frame that has to be swapped out for a context switch. The use of the kernel argument called isolcpus has helped extend the life of these wheezy based installs but apparently does not work on the arm's so we are doomed to use an interface card that offloads the realtime stuff into an fpga card specifically programmed for the job. That adds around $200 to the cost of building a machine because the fpga stuff is 3.3 volts max, and needs a lot of buffering to absorb the noises around one of these machines that would kill an fpga before the first motor to start has moved 5 shaft degrees. I've killed 3 of the $70 cards making a pi run a lathe before I was convinced to buy the 3 buffer cards at $45 each that this particular interface card needed for protection. Zero problems with noise since. But thats a $39 rpi3b, and $200 in the interface. Now I'm in the process of adding a smart breakout card to my 4 axis mill, because a $2 voltage regulator failed shorted, putting 35 volts on a 5 volt breakout board with clean up the exploded epoxy results. So I'm rebuilding that with a 5x more expensive $120 card that will run on 12 or more volts, and also give me enough i/o to build a tool changer and possibly a workpiece pallet changer if I can get them made. But with anything I need. like connectors or cable being a buy it on ebay thing, it will take another month to put the lib back on the box, its done and the machine works again. Keeps me out of the bars don'cha know. ;-) And thats life, no one makes either of those items for any machine at less than 5 to 10 thousand dollars due to the custom labor involved. And do it all on a slow to us, 1 kilohertz control thread. With those, it becomes a write the g-code once, run many times for production, so you can load the machine and turn out the lights and have a bin of finished parts when you turn the lights on the next morning. Thank you, rhkramer -- Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>
