Jacob Repp » unix

Pipe Directories Over SSH

proj — Sat, 12 Jan 2008 17:17:08 +0000

My good friend Jayk showed me this nifty trick years ago for moving directories over ssh:

From a local directory to a remote directory:

tar -zcf - . | ssh name@host "tar -zvxf - -C "

From a remote directory to a local directory:

ssh name@host "tar -zcf - -C  ." | tar -zvxf -

This works by causing tar to read from stdin or write to stdout “-f -” and piping the results through ssh using the ‘execute command over ssh’.

The Unix Haters Handbook

proj — Thu, 29 Mar 2007 22:35:42 +0000

The Unix Haters Handbook was making the rounds on reddit recently. I’ve had it open in a tab and have been occasionaly reading a chapter. I find it very humorous and it’s good to reality check things you take as fact day-to-day. I can agree with a lot of the feelings expressed in the handbook, at I times I have experienced them myself. These days I generally find unix like OSes (especially in the BSD linage) to be a welcome addition to my Macs and Windows environment.

Unix is very good for people who like to understand everything about the machine, who don’t mind reading source code, man pages and command output. I often find myself resorting to ‘unixisms’ in my Windows and Mac environments. This is made easier by the userland FreeBSD binaries on the Mac.

I guess because of learning so much from Unix it has strongly biased me to that type of architecture. I think that there are many strong design paradigms in the various modern Linux and BSD kernels as well as in the userland binaries. Some of these designs may not be executed as completely as on a Symbolics machine for instance and their pervasiveness may frustrate some. These faults only really open up a door of opportunity for anyone who has the insight to improve upon or re-invent what’s there and try to lay a new foundation for the future.

If you fail at least fail trying.

Exploring ‘Load-Average’

proj — Thu, 22 Feb 2007 16:19:42 +0000

This is the best exploration of unix ‘load-average‘ that I have ever read. Most commonly in my experience it has been used as a rough metric based on machine quality of load, which is to say, almost completely worthless metric. The basic idea is that it is a running average of the CPU run queue length. At some point I’m going to have to figure out what the state of the art linux kernel/process profiling tools look like. Hopefully they’re better or at parity with windows/perfmon.