Please wait...
Well, I have started experimenting with piCore, which is a port of Tiny Core Linux to the Raspberry Pi. I am in need of some cheap backup server for hosting the backups of some personal servers of mine and since I had a spare Raspberry and some hard drives , I thought it was worth a try.
I expect crappy performance, but that is accounted for :)
Something I like about Tiny Core Linux is that the OS gets loaded to RAM, and it is small as heck, so it is great for low powered systems and it does not rape the SD card much.
I actually considered NetBSD for the task too, but an operating sistem that loads itself to ram seems so well suited for the task...
Dammit. hit go too fast. Even easier would be to just setup NFS if security isn't a concern.
Mon Mar 22 2021 17:43:36 EDT from Nurb432Armbian includes openmedia vault as a option
2021-03-22 17:44 from Nurb432
Dammit. hit go too fast. Even easier would be to just setup NFS if
security isn't a concern.Mon Mar 22 2021 17:43:36 EDT from Nurb432
Armbian includes openmedia vault as a option
I have thought of NFS.
I am not a great fan of dedicated NAS Operating Systems for things I can deploy myself in 3 minutes. As for NFS, it could do... certainly I am thinking of reducing encryption overhead as not to overwork the rpi (which is an early model) and the clients would be doing the encryption and decryption themselves.
I have been thinking about using FTP because I can upload via FTP using pipers. ie I can do gtar -f - whatever | $encryption_program | curl $upload_instructions. "Whatever" may be a big (for a home environment) pile of data worth 600 GB :)
ANother stupid option I have considered is to use a netcat/ncat solution. I could have the client computer open a control SSH session, and have it use it to instruct the rpi to open a netcat listening port. Then push the whole encrypted file into that port via netcat. However, I think that option is a) Fragile without massive ammounts of scripting and b) overengineered.
I suspect I am going to end up using an ssh pipe with a fast cypher, because it is what the piCore supports out of the box and I doubt it will make a big difference.
2021-03-23 08:14 from Nurb432
If i'm super lazy, SFTP..Tue Mar 23 2021 06:53:38 EDT from darknetuserI have thought of NFS.
This is the third time I try to make my post.
$dump + $gpg + $ssh > $target with a monster pipe configuration do the trick.
$dump + $gpg + $netcat spare a LOT of CPU cycles in the raspberry, and also improve speed somewhat. The drawback is the setup is very frgile (I need a scrypt that connects over ssh to the raspberry and launches a nc server, and then the transfer is initiated, but the gains from using this method do not justify all the error management code this solution needs in order to be somehow solid.
NFS is out of the way because Tiny Core Linux does not have an NFS-utils package for Arm and I don t feel like creating yet another Tiny Core Linux package just for trying some solution and then deciding I don't like it.
Well, seems RMS is back. His cancellation was only temporary.
I call for his immediate removal. Far-left ultra-communists who don't shower and eat their own toe cheese are not suitable for leadership roles.
Armbian does :) and its pretty lightweight.
Wed Mar 24 2021 08:54:33 EDT from darknetuserNFS is out of the way because Tiny Core Linux does not have an NFS-utils package for Arm and I don t feel like creating yet another Tiny Core Linux package just for trying some solution and then deciding I don't like it.
lol
Wed Mar 24 2021 16:40:01 EDT from IGnatius T FoobarWell, seems RMS is back. His cancellation was only temporary.
I call for his immediate removal. Far-left ultra-communists who don't shower and eat their own toe cheese are not suitable for leadership roles.
So i guess everyone is puling funding of the FSF due to his return.
Redhat the latest.
( of course they may jave just been looking for an excuse, without losing face )
speak of the woke supremacists, this account is on fire:
https://twitter.com/WokeRabbi/status/1295342079841837056
Stallman can keep emacs. :)
Ahem,
"Meanwhile, Miguel de Icaza, co-founder of the GNOME desktop project and now an employee of Microsoft, said in a tweet: "[The] 'rms should resign' signatories list contains many significant contributors to free software b people that have had to interact with him and have advanced the cause. The rms support list seems to be mostly users with few credentials b likely fans that never had to deal with him."
I'm sure this just appeared as an illegal string of line noise on IG's screen... ;)
RedHat blasts the decision, says they will be suspending FSF contributions:
https://www.redhat.com/en/blog/red-hat-statement-about-richard-stallmans-return-free-software-foundation-board
EFF says this is a bad thing:
https://www.eff.org/deeplinks/2021/03/statement-re-election-richard-stallman-fsf-board
I'm sure this just appeared as an illegal string of line noise on IG's screen... ;)
"Miguel de Icaza, an employee of Microsoft who they hired to write the GNOME in order to sabotage Linux's best chance of having a standardized desktop..."
FFTFY. :)
RMS has fans in the same way that AOC has fans.
Now watch me piss everyone off (not that it matters)
Today I added a new ethernet interface to a FreeBSD machine. It messed up EVERYTHING. All of the interfaces got renumbered and reassigned, and everything went offline.
This problem was fixed in Linux, quite some time ago, using something I happen to like: SYSTEMD!
The reason Linux doesn't show interface names like "eth0" and "eth1" anymore is because systemd+udev v197 or newer assigns persistent interface names. And it doesn't do it using the older "udev rules" file to try to figure out which is which. Instead, it will assign a name in the following order of priority:
- Names incorporating Firmware/BIOS provided index numbers for on-board devices (example:
eno1) - Names incorporating Firmware/BIOS provided PCI Express hotplug slot index numbers (example:
ens1) - Names incorporating physical/geographical location of the connector of the hardware (example:
enp2s0) - Names incorporating the interfaces's MAC address (example:
enx78e7d1ea46da) - Classic, unpredictable kernel-native ethX naming (example:
eth0)
Thank you, Lennart Poettering, for making sense of something that was previously unpredictable. If the pfSense people used Linux+systemd, I wouldn't have had to waste a bunch of time today mapping out the MAC address of each interface to figure out where they all moved to.