A bit of perl while I’ve been away (Always including some CISCO of course).

Well no matter how many times I head back to post just lately I have never had the time. Studying been put on hold, house move on hold, and working falt out at the new job, but I have to say it has been fun.

Like I say not much studying done, but I have been playing with perl scripting latley. So rather than post nothing I thought I would share the results with you.

At work one of my teams jobs is to provision ports and reclaim ports on switchs. we basicaly write scritps during the day and then batch implement at night. So to make things a bit simpler, (and casue I was bored) I wrote a little web page using perl scripting as the back end.

Now I know I could use client side scripting to get a neater effect, and / or a stand alone program. But one thing about this was I wanted it as client friendly as possible, so decided to go for a web interface. And I have been meaning to learn a scripting language for ages so this seemed an good opportunity.

The web page is in two parts, one for provisioning ports and a second for reclaims, both can handle CATOS and IOS configs

They both still need some tidying up and I want to add some better functions to them soon, but Just so you don’t think i have disappeared here’s the link is you want to try them out. 🙂 (they link to each other)

Port provisioning script

The reason it is set out as it is, is that for my job servers are duel link to two switch, so for each script we are provisioning 3 or 5 ports per server mostly. Also the second switch is normally a mirror of the first, so by ticking the copy check box you only then need to fill in the interface on the second switch, the vlan, speed, duplex, ilo (integrate lights out for use when server crashes to remote reboot and get access to the bios) and sub-net information will be copied from the first switch. Port description as set to the name of the server configured.

I want to put some error checking in the webpage so users can’t skip filling in fields, and then expand to allow users to alter the number of switch / interfaces on the fly.

But for now it seems to work OK, and definitely speeds up or scripting tasks.

Let me know what you think 🙂 If you have any suggestions or would like to see the source code let me know. If you wold like a bit more in depth of what I did may be i can do a follow up artical at some point.

Cheers

DEVILWAH.

PS. little baby is doing great. 9 months old and I finaly understand what people mean when the say how life changing having a baby is! Miss Lilith as she is know in our house is laughing smiling and generaly making my life great. Can’t wait for the warm weather to take her out more 🙂

A Bit More Linux Stuff

While my CCNP studies are going well, this last week I have been setting up some more linux webservers at work, and I came across two issues. Both with simple solutions.

First was installing a SMTP relay server inside the network so websites (such as this one) can send admin  emails. Before I had been using an external smtp server, in the case of this site it was google mail, and while this is great for a single personal site it is not really the way you want to have it set up. One of the main resons of course is that by having your own internal mail server, only that one machine needs to connect to the outside.

It is actualy very easy to set this up, simply installing postfix, giving it a very basic set up. Using unbuntu and aptitude there is even a simple walk through script to get these configurations in.  Below is the default config you need to set up.

myhostname = host.domain.com
alias_maps = hash:/etc/aliases
alias_database = hash:/etc/aliases
myorigin = /etc/mailname
mydestination = domain1.com, domin2.com, localhost
relayhost =
mynetworks = 127.0.0.0/8 [::ffff:127.0.0.0]/104 [::1]/128 192.168.0.0/24 172.16.0.0/16
mailbox_size_limit = 050000000
recipient_delimiter = +
inet_interfaces = all
inet_protocols = all

The important one is “mynetworks = …….”

This is telling postfix that it should act as a rely from any host in these ranges that tries to use it to send mail thought it. But once you have the set up above and as long as the servers are in the ip ranges configured then you are good to go.

Just note a few things! First this is not the most secure set up, any device in the ranges can now send internet mail, this could end up a potential method of attack where malicious software attempts to send spam. Either be very care full in securing what devices can use the relay server or else think about using authentication to secure it.
Secondly if you do not have a static IP, and your mail server does not have a correct public DNS record set up, you may find some companies will refuse to accept mail from it. This is quite a standard way to prevent them getting hit by spam (Zombi clients infected with malware generally don’t have DNS records set up, company mail servers do). So if you expect to be sending lots of mail to a wide variety of internet address so make sure you sort this out.

My next step in this is to extend postfix to not only be a relay for my webservers but to become a full blown email server that revives internet mail for my domains.

The other thing I came across was mod_proxy and mod_proxy_http, A while ago before I knew what I was doing I set up some websites on a server. Now looking back I realise that my first attempt was poor and I need to build a new server up and move the sites over. But with only one external IP to play with and multiply sites there is an issue of trying to keep every thing live and working during this process. The network looks something like below.

And on the old server I have the following virtual host file.

<VirtualHost * >
ServerName  site1.domain.com
# Indexes + Directory Root.
#DirectoryIndex index.html
DocumentRoot /var/XXX/
<Directory /var/XXX/>
Options Indexes FollowSymLinks MultiViews
Options -Indexes
AllowOverride FileInfo
Order allow,deny
allow from all
</Directory>
</VirtualHost>

<VirtualHost * >
ServerName  site2.domain.com
# Indexes + Directory Root.
#DirectoryIndex index.html
DocumentRoot /var/WWW/
<Directory /var/WWW/>
Options Indexes FollowSymLinks MultiViews
Options -Indexes
AllowOverride FileInfo
Order allow,deny
allow from all
</Directory>
</VirtualHost>

So I set the first website up on my new server, but only having one external IP address, I can’t then easily forward external traffic to the new server. NAT/PAT wont work as all traffic is coming in on port 80. However the solution with mod_porxy_http is simple. Once I have set up the site on the new server and configured a virtual host for it. I can then update the virtual host file on the old server for the moved site to contain this entry.

<NameVirtualHost *
<VirtualHost *>
ServerName site1.domain.com
ProxyRequests Off
<Proxy *>
Order deny,allow
Allow from all
</Proxy>
ProxyPass / http://NEW.domain.com/
ProxyPassReverse / http://NEW.domain.com/
</VirtualHost>

Now although the requests are still coming in to the old server it is simply passing them to the new server and relaying the replies back. We can carry on moving each site over in turn until all the old server is doing, is acting as a proxy for all the site that are now running on the new server. We can then update the NAT/PAT if we wish to point directly to the new server and decommission the old server.

Another reason you might want to employ Mod_proxy is if you have several  heavily utilised sites that you want to run on separate hardware, but like this example you only have one public IP address. Having one physical box acting as a proxy server, you can farm out the requests to as many physical boxes as you wish. Taking this further and implementing the server load balancing that comes on many CISCO routers, along with a couple of low spec front end servers acting as proxy to your web-servers behind you can achieve, redundant, load balancing at low cost.

I have often found that set-ups that cost many 10’s thousands, can be emulated with hardware that is often already on site, and a little careful thought.

But that’s been my week so far, fun stuff 🙂

DevilWAH

Do you like the Pretty links?

Getting the pretty Permalink’s to work on this blog has been a bit of a pain, According to word press you click on the format you want under the settings and then they should all work nicely.

so rather than have a link that looks like

“http://www.devilwah.com/?p=344”

you can have the same link looking like

“http://www.devilwah.com/2010/09/minority-report-the-reality/”

Pretty 🙂 right.. 😉

So how did I get it all up and running?

I found out when first trying to activate it that I came across a “page not found error” suggesting that the mod_rewrite module in Apache was not running correctly. And after lots of searching around I found it this is to do with the “Allowoveride” directive in Apache.

The default setting for the directive in the virtual site file in Unbuntu is,

<Directory />
Options FollowSymLinks
AllowOverride None
</Directory>

Further reading suggests that with this set to none, the .htaccess file that is needed for  mod_rewrite to be able to work will not be used.

Searching the net lots of people suggest changing this to “Allowoverride All”, which after a restart of Apache will work fine. But for a little more security I found “Allowoverride FileInfo” will achieve the same thing.

And that’s it, one little word change is the difference between it all working fine and page not found!

The same can be achieved by editing the httpd.conf and associated config files, but as I use virtual sites I prefer editing these directly.

Thank fully the old style links still work just find, the mod_rewrite simple takes the pretty version of the link and translates it back to the ugly version behind the scene. Leaving you the user with a more pleasurable browsing experience.  🙂

DevilWAH