John Siracusa's questions

What is the simplest way to configure apache 2.2 to listen on, say, 20 different ports and then proxy those requests to the same ports on another server? For example, a request to http://myserver.com:1000/ should be proxy-forwarded to http://otherserver.com:1000/, a request to http://myserver.com:1001/ should be proxy-forwarded to http://otherserver.com:1001/, and so on, for ports 1000 through 1019.

Getting my server to listen on 20 different ports is easily done with 20 Listen directives. The trouble is the proxy directives. The mod_proxy module's ProxyPass directive doesn't seem to have a place for any reference to the port number. I can make 20 different VirtualHost sections and put proxy directives in each one:

<VirtualHost *:1000>
  ProxyPass        / http://otherserver.com:1000/
  ProxyPassReverse / http://otherserver.com:1000/
</VirtualHost>

<VirtualHost *:1001>
  ProxyPass        / http://otherserver.com:1001/
  ProxyPassReverse / http://otherserver.com:1001/
</VirtualHost>

<VirtualHost *:1002>
  ProxyPass        / http://otherserver.com:1002/
  ProxyPassReverse / http://otherserver.com:1002/
</VirtualHost>

...

but that seems extremely tedious. Though I can generate all those blocks programmatically from a template, I have to believe there's a better way, maybe with a single mod_rewrite rule or something? Suggestions?

I have an apache server whose main purpose is to handle RPC requests. All such requests hit a url that begins with a fixed prefix (e.g., /rpc) I use a Location directive to set the handler for this URL (e.g., <Location /rpc>...</Location>).

I'd now like this server to also serve static files. I put an index.html file into its DocumentRoot but cannot get apache to serve it to me. Instead, I get 404 responses. The access log shows this:

127.0.0.1 - - [18/Aug/2010:11:41:52 -0400] "GET /index.html HTTP/1.1" 404 208

The error log shows this:

[...] [error] [client 127.0.0.1] File does not exist: /www/htdocs/index.html

This error message is a filthy lie. I can copy and paste that path and run ls /www/htdocs/index.html and see that the file does indeed exist.

Things I have tried so far:

• Checked all permissions in full path to the file: all dirs are 755, all files are 644
• Checked ownership: all dirs and files are owned by the same user that apache is running as
• Set LogLevel debug (no new messages appeared in the logs)
• Added a <Location /> block with a SetHandler default-handler directive
• Added a <Location /index.html> block with a SetHandler default-handler directive
• Moved both of these Location blocks above and below the <Location /rpc> block
• Ran dtruss and did not see apache issue a single stat() call while processing the request

I'm running out of ideas. What else should I try?

(I'm using apache 2.2.12 on Mac OS X 10.5.8 and apache 2.2.3 on CentOS 5.4.)

Consider these two Puppet module files:

# File modules/a/manifests/b/c.pp
class a::b::c {
    include b::c
}

# File modules/b/manifests/c.pp
class b::c {
    notify { "In b::c": }
}

It seems that when Puppet hits the include b::c directive in class a::b::c, it searches for the corresponding *.pp file by looking backwards from the current class and decides that it find the correct file located at ../../b/c.pp. In other words, it resolves b::c to the same *.pp file that the include b::c statement appears in: modules/a/manifests/b/c.pp

I expected it (and would like it) to instead find and load the file modules/b/manifests/c.pp. Is there a way to make Puppet do this? If not, it seems to me that module names may not contain any other module names anywhere within them, which is a pretty surprising restriction.

Our entire organization requires PGP Whole Disk Encryption (WDE) to be used on all laptop and desktop computers. For Macs running Mac OS X 10.5 or later that are using Time Machine, the (local) Time Machine backup disk must also be encrypted.

One of the promises of Time Machine is that, in the event that your boot drive fails, you can replace it with a new, empty disk, boot from the DVD that came with the Mac, and restore the boot volume contents from the Time Machine backup disk onto the empty disk.

Unfortunately, in our case, since the Time Machine backup disk is also encrypted using PGP WDE, the boot DVD cannot read it and therefore cannot recover data from it. There are several obvious sloutions that I've read about on the web, including:

• Create a bootable DVD or hard disk that is the same as the DVD that came with the Mac, except that it also has PGP WDE installed.

• Boot the Mac that has the failed drive in FireWire target disk mode, then connect it to another Mac that has PGP WDE installed and use that Mac to decrypt the Time Machine volume. Then follow the usual DVD-based Time Machine recovery procedure.

These both sound like they should work. What I'm looking for is a testimonial from someone who has actually done it at least once, using one of these techniques or another one, and seen it work. Please also provide as much detail as you can for the technique that you used. (E.g., if you created a bootable DVD with PGP WDE installed, what software did you use to make it?)