Going to SCALE 8x today - I'll be in the LOPSA booth, PostgreSQL booth and doing FOSS tours. Do I know how to overcommit or what?

If you have multiple vias in your rsync path (e.g. a -> b -> c) you
can use an rsync filter rule to tell rsync: don't transfer files that
have . in the beginning of the filename!

--filter '- .*'

This way you won't have rsync temp files (named something like
.filename.xxyyzz) from the a -> b transfer ending up on c.

http://sysadvent.blogspot.com/

Just submitted an article on Cfengine 3 to http://sysadvent.blogspot.com/ - I guess it'll be a few days before it's posted.

After much tearing out of hair, learned to use dblatex (not passivetex) to generate PDF's from my DocBook XML. Runs faster and smoother - that is to say it actually generates the PDF instead of just spewing out LaTeX errors.

I'm trying AsciiDoc to obviate the XML "tag soup" I am eating with my DocBook docs.

There were two things I had to tweak to get the asciidoc .TXT -> DocBook XML -> PDF pipeline working on CentOS 5.3:

1. Remove or comment out every mention of the "--nonet" switch in my xmlto (v0.0.23) (xmlto is the wrapper script for the DocBook XML -> PDF process). AsciiDoc generates DocBook XML 4.5, and the CentOS docbook-dtds package only goes up to 4.4, so we need to be able to access the DocBook XML 4.5 DTD over the net.

2. Do NOT put any author info into the AsciiDoc file, as leaving it in would break the LaTeX leg of the process. (DocBook XML -> XHTML would work fine, but DocBook XML -> PDF (or DVI) would break.) Once I took that out, I was able to generate PDF files.

This is not the ONLY or even the BEST way to do it - this is just my way.

I took 10 years to travel this path, and I made many friends along the way.

In a nutshell, I took about 7 years of routinely working 12 hour days, plus 2 years practicing dealing with people and learning about people, and then back to computing again.

Here is the long version:

I started with 2 years of part-time work as a help desk technician at a private university while enrolled as a student (about 15-20 hours per week).

Upon entering the real world in 1995, joined the help desk at a rapidly growing ISP (EarthLink).

Download xtrlock source code from the Debian project (I used http://ftp.debian.org/debian/pool/main/x/xtrlock/xtrlock_2.0-13.tar.gz)

Follow the instructions in debian/README.Debian:

xmkmf

make

Then, take the first of the two gcc commands that were run by make, and rerun it manually, but add -DSHADOW_PWD to the defines section of the command line (-Dthis -Dthat -DSHADOW_PWD). (I tried to muck around with the 1000+ line Makefile but found it easier to just re-run the gcc commands)

Then re-run the rest of the commands from that previous run of make

make install

chmod u+s /usr/bin/xtrlock (if you want non-root users to use it)

I posted my cfengine config, it's the 2nd half of my cfengine presentation to UUASC-LA (chapters 14 - 34).

http://www.verticalsysadmin.com/cfengine/cfengine-presentation-to-uuasc-la-on-1-oct-2009/

Maybe it can help somebody start or expand their cfengine config.

Presenting on cfengine 2 made me really excited to learn 3... as it's promised that the cfengine language has been standardized and streamlined in 3, and it's the lack of that in 2 that has turned some people away. I only perservered by picking the "shellcommands" modules as my favorite module and writing lots of shell scripts that are run and administered by cfengine.

http://blog.sqawasmi.com/index.php/2009/09/28/ping-tip-ctl-backslash/

confirmed on Linux