On 3/26/22 17:20, DdB wrote: > Model: Samsung SSD 970 EVO Plus 500GB > ... DUAL CPU's (32 cores in total without HT), 128 GB ECC-RAM, > 1 NVME SSD + 8 Hotplug HDD's, used for plenty of virtualized machines. On 3/26/22 19:50, DdB wrote:
Hey, excellent! much to my liking, you are coming up with most of the things, i already do. But never stop improving on them ... Yes, i use VCS, in my case git. Yes, i use SSD, in my case nvme, yes, i use SSH (even from a vm to the host), in order to create a simulation of it for experimentation. (undocumented, but very handy). BTW: That is one of the reasons to upgrade, bcoz virtualbox is no longer maintained proper on buster. Of course, i use snapshots and zfs-auto-snapshot and many services on top of it. I even do keep images of the OS in the pool for reference, history or for having the option to rollback to it. Ok, the thing with the UIDs is going to be useful, as i maintain those manually uptil now. The only step, i am refusing to take is the replacing of SSD's (I couldnt do that on my own and would require assistance always.) Thus i prefer to multi-boot, which is fine, as long, as you know, what you are doing and keep most of its config static, which i do. Also, i did avoid SMR since years, and i have no need for samba at all. Everything is local (or local to a vm). I was using a fast SSD for ZFS, but theat one got consumed due to all the IO from the cache device. Now, i opted for large enough RAM, which gives even better performance. As you can see, we are almost on the same page, and i might add: i am enjoying that a lot!
That is a powerful computer.I used to create unprivileged and system accounts manually, until I wrote a script and a data file to do it for me.
Multi-boot "dangerous installations", bootloader damage, etc., are what drove me to the drive rack and one-OS-per-disk solution.
To migrate to a new OS, I backup, power down, pull the old disk, insert a wiped disk, boot the installer, do a fresh install, reboot into the new OS, and restore. If the new OS does not meet my expectations, I revert to the old disk. As the old OS was untouched by the failed upgrade, all I have to worry about is whatever data changed during the attempt (which should be limited to e-mail client files, because I use a VCS and a file server).
My daily driver is a Dell Latitude E6520 with an Intel Core i7-2720QM processor, 2 @ 8 GB memory, Intel 520 Series SSD 60 GB, and NVIDIA Optimus graphics. The graphics have always been a problem on FOSS OS's. New major versions typically fail -- e.g. Debian 9.0, 10.0, and 11.0. When this happens, I put the new disk on the shelf and continue to use the old Debian. A month later, I update/ upgrade the new Debian in a different machine and try again. When the new Debian seems to work correctly in the laptop, I keep it and the migration is complete. Migrating from Debian 7 to Debian 9 was delayed until around version 9.5 (?). Debian 10 has never worked right. Debian 11 was close the last time I tried, but I am still on Debian 9.13 (which has gotten worse).
David