Recursive find and replace

From time to time I find myself wanting to replace a bit of text, in multiple files, throughout a huge directory tree (particularly in the case of a Wordpress migration).

Here’s a method I frequently use (other examples I’ve seen out there make use of perl, xargs, grep, etc.):

user@host:~$ cd Directory_To_Start_From
user@host:~$ find . -type f -exec sed -i 's/Text_To_Find/Replacement_Text/g' {} \;

The find command will return a list of all the files in the directory tree and execute the sed command on each one it locates. In the example below, I added the “-name” option to the find command to allow for working only on files with the .js extension:

user@host:~$ cd website
user@host:~$ find . -type f -name "*.js" -exec sed -i 's/var pageName="example page";/var pageName="Final Page";/g' {} \;

In the next one, the text to find and replace has special characters (in this case the ‘/’ in the URL).

user@host:~$ cd website
user@host:~$ find . -type f -exec sed -i 's/http:\/\/www.example.com\/old_directory\//http:\/\/newsubdomain.example.com\/new_directory\//g' {} \;

See the following links to read more about escaping special characters:

Posted by Karl Herrick on February 8th, 2010 in Wordpress, Linux, Hosting, Bash | No Comments »

Sloppiness at 1and1

So I have been noticing issues with 1and1 again. Yesterday when trying to get into my hosting control panel at admin.and1.com, I received a “HTTP Error 503 Service unavailable” error, and thought to myself… really, on the admin panel… really?

Today while checking out my own site, I noticed that all of the PHP files I tried browsing to returned “500 Internal Server Error” errors.

Now this has gone too far… so I called them up. At first the person I talked to explained that I needed to force PHP5 using a .htaccess file, and proceeded to tell me that they could make the change for me. Calmly, I responded by saying that I did not want a change, and that it had been working fine the way it was for months.

She put me on hold, and while I waited… it started working. She told me that the shared hosting server “had a glitch”.

Afterwards, I ssh’ed into my server. Look what I found:

(uiserver):uXXXXXXXX:~ > ls -al
total XX
-rw-r–r–  1 uXXXXXXXX XXXXXXXX   16 Feb  3 14:06 info.php

I thought, “weird, I didn’t put that there”.

(uiserver):uXXXXXXXX:~ > cat info.php
<? phpinfo(); ?>

Shame on you 1and1, for not cleaning up after yourself. Well, I guess you get what you pay for.

Posted by Karl Herrick on February 3rd, 2010 in Culture, Linux, Hosting, Bash | No Comments »

Interesting Analytics

I thought this would be fun to show everyone.

Accident Analytics

The amount of visitors coming to karlherrick.com is fairly consistent, so I was surprised to see such a spike when I was in a car accident recently. There must be some correlation to draw here.

I am thinking of breaking my leg next week to drive the traffic back up. ;-)

Just kidding.

Posted by Karl Herrick on January 19th, 2010 in Web Development | No Comments »

Switching MySQL Databases on 1and1

After wanting to test Wordpress 3.0 on a 1and1 hosting account, I ran into the following issue… the particular setup I was dealing with had only one database available, and it wasn’t MySQL 5, but instead MySQL 4… furthermore, it was a version lower than 4.1.2, which more recent versions of Wordpress require as a minimum.

So after doing a bit of research on what others had done, I went about fixing the problem. Interestingly enough, 1and1 could have made some money in this situation if they would allow customers to simply purchase another database to add to their packages, but no, they want an entire upgrade to be purchased.

The first step was to backup the website and database. I ssh’ed into the web host:

username@localhostname:~$ ssh $username@example.com

Then I backed up the database. The command below will dump all of the databases on the host (in this case, only one MySQL 4.x database) into a dated bzip file, in the home directory.

username@remotehostname:~$ mysqldump -C -A -u $databaseUsername -h $databaseHostName -p$databasePassword | bzip2 -cq9 > ~/`date +%F-%I-%M-%p`-db-backup.sql.bz2

Now to backup the site itself (not only in case disaster struck, but this would get a local copy of the sql dump I just made as well). On a local OS X workstation (or Ubuntu, if that suits your tastes) I ran something similar to the following:

username@localhostname:~$ mkdir ~/website_backups
username@localhostname:~$ rsync -avz –exclude="logs" $username@example.com: ~/website_backups/

From there, I was able to log into the 1and1 control panel and delete the existing database. This allowed me to setup a new one, and in particular, choose MySQL 4 or 5 as the type.

Back to the 1and1 hosting account:

username@localhostname:~$ ssh $username@example.com

I uncompressed the sql dump:

username@remotehostname:~$ bunzip2 2010-01-14-01-59-PM-db-backup.sql.bz2

and was greeted with the raw sql in the file, “2010-01-14-01-59-PM-db-backup.sql”. From here it was only a hop skip and a jump away to restoration. It was necessary to edit the sql file in order to have it restore properly to the newly created database that was just created:

username@remotehostname:~$ nano -w 2010-01-14-01-59-PM-db-backup.sql

And I changed the $oldDatabaseName to $newDatabaseName.

–
– Current Database: `$oldDatabaseName`
–

CREATE DATABASE /*!32312 IF NOT EXISTS*/ `$oldDatabaseName`;

USE `$oldDatabaseName`;

ctrl+o, enter, and ctrl+x, to save the file and exit nano. The sql dump was now ready to restore to the new database.

username@remotehostname:~$ mysql -u $databaseUsername -h $databaseHostName -p$databasePassword $databaseName < 2010-01-14-01-59-PM-db-backup.sql

The only other things to do were to update any existing applications that needed the new database name, username, hostname, and password.

As a note, if you are updating a Wordpress install to point to a new database, this info can be changed in the file, wp-config.php.

Posted by Karl Herrick on January 14th, 2010 in Wordpress, Backups, Apple, Linux, Hosting, Web Development, Bash | No Comments »

WDS Bridging Experiences

At the moment, our best solution for network coverage to multiple buildings is wireless. I have been planning out various solutions to improve our system (fiber optics, dslam, or a better wireless setup) but wanted to share what we have now and how well it has worked in this particular area of our property.

WDS Repeating Map

The image above shows the main buildings that are being linked. The main router sits in an upstairs window and is a:

Linksys WRT300N

The three client WDS routers are also sitting in windows, and are on the first level of the buildings that they are in. They repeat the same SSID, and run on channel 11.

Buffalo WLI-TX4-G54HP

(2) Belkin F5D7230-4 routers

This setup has been going for weeks without an issue, and has served upwards of six families, a two person office, a fifteen to twenty person office, and a classroom with sixteen students.

We have other areas on the same property that have wireless access between buildings, but are nowhere near as complete as this arrangement. If we do end up using wireless to unify the property’s networks, then I will start looking into site to site links, multiple gateways, OLSR, and other mesh technologies. I have my doubts as to how well WDS scales. Has anyone had positive results with professional mesh solutions built on DD-WRT and Linksys hardware?

Posted by Karl Herrick on January 7th, 2010 in Wireless, Linux | No Comments »