PMA Media Group» IT

Database syncronization woes

Narshlob — Wed, 04 Nov 2009 00:00:50 +0000

Database resyncronization depends on what went wrong but the steps below will most likely solve most issues.
Run these commands on the slave database

STOP SLAVE; # stop the Slave I/O threads
RESET SLAVE; # forget about all the relay log files

Then go to the master database and run these

RESET MASTER; # reset the bin log counter and wipe out bin log files
FLUSH TABLES WITH READ LOCK; # flush buffers and LOCK tables
show master status\G

Note what the show master status command returns. You’ll need to know the file name and the position.
You can do one of two things here, make a dump of the entire master database (in which case I suggest you follow this)
or you can just update the tables.
Usually we just need to update the tables so release the lock on the master database tables (UNLOCK TABLES;) and then run this command on the slave database (download maatkit tools here),

cd ~/maatkit-5014/bin && sudo ./mk-table-sync –[print][execute] u=[user],p=[pass],h=[master_host_name] –databases [database_name(s)] localhost

I suggest you run –print before you run –execute. If you run –execute first, you have no idea what just happened. –print will let you know what it’ll do without actually doing anything.
Back to the slave database mysql client, issue these commands,

CHANGE MASTER TO MASTER_LOG_FILE=’[file name from show master status command]‘, MASTER_LOG_POS=[pos];
SLAVE START;

Run this command,

show slave status\G

And check that these aren’t NO or NULL,

Slave_IO_Running: Yes
Slave_SQL_Running: Yes
….
Seconds_Behind_Master: 1634

If things aren’t back to normal, follow the instructions on this website.

Working with Git

Narshlob — Fri, 19 Jun 2009 23:34:48 +0000

This tutorial covers all the commands (hopefully) we’ll need for the projects we build here at PMA. If there’s anything that needs to be added to it, feel free to comment.

Starting with the basics, we’ll first cover retrieving a project:

git clone [repository]

This will get the currently active branch from the repository

Obviously you’ll want to do something with this newly retrieved working copy of the project. Let’s first create a branch for the new features/bug fixes we’ll be coding.

git checkout -b [newbranchname] origin

Ok, so we got a new branch. While we’re coding, it’s a good idea to commit tons of times to preserve the changes we’ve just made. Don’t worry about the log, we’ll make it pretty later. Just commit often.

git add [filename(s)]
git commit

If it’s a small change that doesn’t require much explanation, you can use these commands

git add *
git commit -m "The commit message"

Or, even shorter

git commit -a -m "The commit message"

You finished that feature so now it’s time to merge that branch with the master branch (or some other branch depending on what VCS (Version Control System) paradigm you/your team chose). First, you should make those hard-to-read commits less hard-to-read. Let’s rebase!
Please note that if you rebase after pushing to the repository, you will create problems for those pulling from that repository. Rebase changes the history of the project. Your teammates merges will not be fast-forward[able]. It won’t be pretty, trust me.

git rebase -i HEAD~[number of commits back]

You’ll now be looking at something similar to this:

pick ce86448 A random commit
pick a8564a9 Another random commit

How you order things in this editor will affect the order of the commits. Note that merge commits are not shown here. They aren’t editable.
Replacing “pick” with “edit” will allow you to edit the changes you made as well as the commit message.
After you’ve edited the files you wanted to edit, you can now

git add *

then

git commit --amend

and move on.

As I mentioned before, you have the opportunity to clean up the mess you made with all those many commits using the rebase option. Here’s how:

Run the
```
git rebase -i HEAD~[x]
```
command from earlier

Replace “pick” with “squash” on the commit you want to be combined with one exactly previous to it.

pick ace72dd I squashed these commits. I'm cool.
squash e99fd59 This commit will be sqaushed with the one above it
pick d0770e8 commited again
pick af845d0 I'm really committed

Pretty straight forward and easy. Everyone loves rebasing
If you want to know more about git-rebase, I recommend checking it out here.

Now, you’ve made all these changes and everything looks great. What are you gonna call this pretty new feature? Are you gonna tag it? I would..

git tag -a [fancy_feature] -m "A fancy message for a fancy feature"

If you happen to leave out the -m, git will open an editor and you’ll be able to add your fancy message there, just like with commit!
Read more about tags here.
One scenario for using tags could be that, within a project, one wants to keep track of versions. Each commit could be tagged with a version number like with bug fixes in some VCSs. If something goes wrong, it’s really simple to go back to a previous version using git rebase, as we’ve already shown.

You’ve squashed those ugly commits, changed the commit message(s), and tagged everything. Time to merge. Switch to the master branch (or whatever branch you’re wanting to merge with) and type this command which will merge [branchname] with the current branch:

git merge [branchname]

You’ll probably want to fix any conflicts and continue with the merge.
It’s now safe to delete [branchname] because all the changes from that branch are now on the current one.

git branch -d [branchname]

This next feature is pretty neat. Say you’ve done a bunch of changes that haven’t been committed yet and you realize you aren’t on a feature branch. Here’s what you do:

git stash
git stash branch [newbranchname]

This will stash away all uncommitted changes, create a new branch and check it out, then unstash all your stashed changes into that new branch. Awesome

Stash is also useful in scenarios where you don’t want to commit yet but you need to switch to a different branch and do something. You could stash the current changes using the above git stash command, do the needed changes, then switch back to the branch you were working on and use this command to unstash the changes:

git stash apply

It’s a good idea to check out the other things offered by git stash (git-stash)

To get this new branch into the origin repository, do:

git push origin [branchname]

To delete a branch from the origin repository, do:

git push origin :[branchname]

Don’t forget the colon!

Another scenario; your co-worker does some work on a feature and gets stuck. You don’t want to type on their computer cause yours is set up just the way you want it. Is there a solution to this quandary? Yeah. There is…
Tell them to push their changes then do this:

git fetch origin [remotebranchname]:[localbranchname]

You now have the branch they were working on locally and can modify to your hearts content.

Noteworthy Notes

There’s a difference between

git pull

and

git fetch

The difference is that “git pull” will run a “git fetch” then a “git merge” to merge the retrieved head into the current branch.

git log -p path/to/file.rb

This command will show the history of a specific file;

git blame path/to/file.rb

will go line by line in a file and give a short description + the name of the person that changed the line last (brilliant, actually)

From the Git manual

check it out
A few configuration variables (see git-config(1)) can make it easy to push both branches to your public tree. (See the section called “Setting up a public repository”.)

$ cat &gt;&gt; .git/config &lt;
      [remote "mytree"]
           url =  master.kernel.org:/pub/scm/linux/kernel/git/aegl/linux-2.6.git
           push = release
           push = test
EOF

Then you can push both the test and release trees using git-push(1):

git push mytree

or push just one of the test and release branches using:

git push mytree test

git push mytree release

To rebase your current working tree to obtain the changes from the master tree,
Suppose that you create a branch “mywork” on a remote-tracking branch “origin”, and create some commits on top of it:

git checkout -b mywork origin
$vi file.txt
$ git commit
$ vi otherfile.txt
$ git commit

…
You have performed no merges into mywork, so it is just a simple linear sequence of patches on top of “origin”:
o–o–o <-- origin
\
o--o--o <-- mywork
Some more interesting work has been done in the upstream project, and "origin" has advanced:
o--o--O--o--o--o <-- origin
\
a--b--c <-- mywork
At this point, you could use "pull" to merge your changes back in; the result would create a new merge commit, like this:
o--o--O--o--o--o <-- origin
\ \
a--b--c--m <-- mywork
However, if you prefer to keep the history in mywork a simple series of commits without any merges, you may instead choose to use git-rebase(1):

$ git checkout mywork
$ git rebase origin

This will remove each of your commits from mywork, temporarily saving them as patches (in a directory named “.git/rebase-apply”), update mywork to point at the latest version of origin, then apply each of the saved patches to the new mywork. The result will look like:
o–o–O–o–o–o <-- origin
\
a'--b'--c' <-- mywork
In the process, it may discover conflicts. In that case it will stop and allow you to fix the conflicts; after fixing conflicts, use git add to update the index with those contents, and then, instead of running git commit, just run

$ git rebase --continue

and git will continue applying the rest of the patches.
At any point you may use the —abort option to abort this process and return mywork to the state it had before you started the rebase:

The commands

Here’s a list of all commands covered in this tutorial:
git-clone
git-checkout
git-add
git-commit
git-rebase
git-tag
git-merge
git-branch
git-stash
git-push
git-pull
git-fetch

Advanced Routing Wireless To Your LAN With Cisco WRT400N

Aaron Murphy — Wed, 10 Jun 2009 23:16:28 +0000

Advanced routing wireless to LAN with the Cisco WRT400N router is simple, once you know how. Just do the basic setup on the router as usual, only disable the DHCP server. Then go to the Advaned Routing page and disable NAT and enable Dynamic Routing. Now you can connect a local network to one of the four LAN ports. Don’t use the WAN port as the Internet connection will go through your main LAN.

Setting PowerDNS To Ignore Records For Downed Web Sites

Aaron Murphy — Wed, 10 Jun 2009 23:15:03 +0000

If you are using PowerDNS for round robin on multiple websites, you can set it up so that it will return only the records for sites that are up. I set up a Ruby daemon to monitor sites and connect to the PowerDNS MySQL table used by PowerDNS on Rails. You can use any language or system you like. It just needs to be able to access your PowerDNS database.

First add the following line to /etc/powerdns/pdns.conf:

 gmysql-any-query=select content,ttl,prio,type,domain_id,name from records where name='%s' and (prio<>-1 or prio is null)

Thats the trickiest part. If you mess up on the sql query, PowerDNS may not return any results for your domain.
Then restart pdns. I use CentOS so I type

service pdns restart

Now you just need a daemon to monitor your websites. Something like the following might work for you:

#!/usr/bin/env ruby
require 'net/http'
require 'rubygems'
require 'activerecord'
 
SITE_ADDRESSES = %w( www.yourdomain.com www.example.com )
SITE_THAT_IS_ALWAYS_UP = "" # some well known site that is always up
 
class Record < ActiveRecord::Base
  ActiveRecord::Base.establish_connection(:adapter => "mysql",
    :encoding => "utf8", :database => "powerdns_production",
    :username => "your_mysql_username", :password => "your_secret_password_here",
    :host => "ns1.your_isp.com")
  set_inheritance_column :ruby_type
end
 
$running = true
 
def signal_trap_exit(signal_name)
  puts "#{Time.now} received #{signal_name} signal"
  $running = false
end
 
our_exit_signals = ["TERM", "INT", "KILL", "HUP"]
for exit_signal_name in our_exit_signals
  Signal.trap(exit_signal_name, "signal_trap_exit('#{exit_signal_name}')")
end
 
def check_internet
  begin
    Timeout::timeout(5) do
      Net::HTTP.get_response(SITE_THAT_IS_ALWAYS_UP, "/", 80) do |response|
        return response.code.match(/2|3\d{2}/)
      end
    end
  rescue => e
    puts "#{Time.now} Error connecting to internet: #{e}"
    return false
  end
end
 
def get_site_ip_addresses
  begin
    Record.find(:all, :conditions => ["type = ? and name in (?)", "A", SITE_ADDRESSES])
  rescue => e
    puts "#{Time.now} Error finding all records: #{e}"
    return []
  end
end
 
def make_inactive(ip_record)
  Record.update_all("prio=-1", "name='#{ip_record.name}' and content='#{ip_record.content}'")
end
 
def make_active(ip_record)
  Record.update_all("prio=1", "name='#{ip_record.name}' and content='#{ip_record.content}'")
end
 
while($running) do
  get_site_ip_addresses.each do |ip_address|
    begin
      Timeout::timeout(5) do
        Net::HTTP.get_response(ip_address.content.to_s, "/", 80) do |response|
          puts "#{Time.now} Response code for #{ip_address.content.to_s}: #{response.code}"
          make_active(ip_address) if ip_address.prio == -1
        end
      end
    rescue => e
      puts "#{Time.now} Error connecting to #{ip_address.content.to_s}: #{e}"
      if check_internet
        make_inactive(ip_address) unless ip_address.prio == -1
      end
    end
  end
 
  sleep 15 if $running
end

How to do Benchmarking with Ruby

Alan Carl Mitchell — Thu, 21 May 2009 16:09:01 +0000

Benchmarking with Ruby is super easy. There is already a built in class–Benchmark–that will do all of the heavy lifting for you.

If you want to do basic benchmarking, Benchmark.bm is the easiest way to go. Take a look below. Here we are going to test three ways to do looping in Ruby and see which one we like best.

require 'benchmark'
 
n = 5000000
 
Benchmark.bm do |x|
  x.report("for loop:")   { for i in 1..n; a = "1"; end }
  x.report("times:")      { n.times do   ; a = "1"; end }
  x.report("upto:")       { 1.upto(n) do ; a = "1"; end }
end

This will produce output something like this:

               user     system      total        real
for loop:  0.727000   0.000000   0.727000 (  0.727030)
times:     0.571000   0.000000   0.571000 (  0.571588)
upto:      0.523000   0.000000   0.523000 (  0.522947)

If we use Benchmark.bmbm, then it will do a ‘rehearsal’ run first in order to better equalize the garbage collection environment of the code inside the block so that hopefully we get more realistic timings.

require 'benchmark'
 
n = 5000000
 
Benchmark.bmbm do |x|
  x.report("for loop:")   { for i in 1..n; a = "1"; end }
  x.report("times:")      { n.times do   ; a = "1"; end }
  x.report("upto:")       { 1.upto(n) do ; a = "1"; end }
end

This will produce output something like this:

Rehearsal ---------------------------------------------
for loop:   0.756000   0.000000   0.756000 (  0.756298)
times:      0.477000   0.000000   0.477000 (  0.476855)
upto:       0.527000   0.000000   0.527000 (  0.526522)
------------------------------------ total: 1.760000sec

                user     system      total        real
for loop:   0.751000   0.000000   0.751000 (  0.751067)
times:      0.516000   0.000000   0.516000 (  0.515874)
upto:       0.581000   0.000000   0.581000 (  0.581254)

Alternately, you can use the Benchmark.realtime method like this:

puts "for loop: #{Benchmark.realtime {for i in 1..n ; a = "1" ; end}}"

and get something like this:

for loop: 0.682438850402832

The Fundamental Rule of Computing

jaredd — Mon, 13 Apr 2009 20:04:14 +0000

I have created my own fundamental rule of computing. And I call it “The Fundamental Rule of Computing”
And it consists of three parts:

Backup everything.
Backup everything to an offsite location.
Backup everything to another offsite location.

PowerDNS With Empty CNAME = Fail!

Aaron Murphy — Tue, 07 Apr 2009 20:58:26 +0000

Just used up my morning trying to resolve an MX record problem. We were not getting our email through our new Google Apps account. I tried to send from Gmail to myself and it worked. I tried to send to someone else in our domain, and got a “550 550 Unrouteable address (state 14)” error.

I tried to dig for our domain. It didn’t report our correct MX record. We use PowerDNS on Rails. This makes it easy to administer PowerDNS. The settings for the Google MX records were right.

It turns out that earlier we had changed our domain to a new cloud system. We had to setup CNAME records in our DNS so that users would get routed properly. Luckily, we use DNS Stuff to help us out. They report “RFC1912 2.4 and RFC2181 10.3 state that there should be no CNAMEs if an NS (or any other) record is present.”

So I removed the CNAME record and everything works now.