05 Mar 2020

Clustered Web Applications - Mysql and File replication

In mid-2018, one of our clients asked if we could improve the reliability of their web applications. The system was developed by us and was hosted on a single server in Hong Kong. Over the last 5 years or so, the server had been sporadically unavailable due to various reasons

  • DDOS attack on the Hosting provider's network
  • Hardware failure - both on the hosting machine and the provider's network hardware.
  • Disk capacity issues

While most of these had been dealt with reasonably promptly, the service provided by our client to their customers had been down for periods up to a day. So we started the investigation into the solution to make this redundant and considerably more reliable.

Since this was not a financial institution, with endless money to throw at the problem, Amazon, Azure etc. were considered to pricey, and even if they did provide a more reliable solution, there was still a chance that it could still be susceptible to network or DDOS attacks. So the approach we took was to build a cluster of reasonably priced servers (both physical and virtual) hosted at multiple hosting providers.

This represented the starting point, we had already separated the Application and Mysql server into individual containers. Which made backups and restoration trivial, along with theoretically making the cluster implementation somewhat simpler

To implement a full clustering solution, not a redundancy solution, we needed to solve a few issues

  • Mysql Clustering
  • File system Clustering
  • Load Balancing
  • Private Networking between the various components.

The simplest of these was the Load balancing, we had already been using Cloudflare to provide free SSL (we tend to use letsencrypt on solutions these days, but Cloudflare has proved reasonably resilient. although it does still result in a single point of failure from our perspective)

The other two however proved to be more challenging than we expected.

Mysql Clustering

Anyone who has used MySQL, normally at some point set's up a master/slave backup system. It's pretty reliable, however, when it comes to switching from the master/slave, we concluded that the effort involved, especially considering the size of our database would be problematic. So we started testing out the Mysql Clustering technologies (note we tended to stick to classic MySQL technologies, rather than trying out any of the forks/offshoots). 

After our initial analysis, we settled on NDB clustering, the setup of which proved more than a little problematic. In part due to the database restrictions that the storage engine enforced, but eventually having overcome the initial issues with this, by modifying our schemas slightly, we discovered that in our usage scenario, that NDB performance was significantly slower than that of a standalone InnoDB server. To the point where the application became un-usable. This may have been due to various factors, memory limitations, one of the machines using a physical rather than SSD drive. But after many hours of research and testing, we concluded that it was not a viable solution.

After throwing all that research in the bin, the next alternative was an InnoDB cluster. Again this involved quite a learning curve as management of the cluster is done via mysqlsh, which due to the nature of the internet has a wealth of out of date contradicting information all over the internet. Along with rather limited precise information on working configurations. Eventually, we managed to solve both the multitude of configuration settings (enough memory allocated to migrate) and minor schema modifications to enable replication to work. Resulting in the first part of the puzzle being solved.

The final solution for the mysql server involved hosting on 1 physical machine, one virtual machine in Hong Kong and a Linode VPS in Singapore. This has generally met the initial goals of more stability, however, we do have a long term plan to move more to Linode, and remove the Hong Kong physical hardware, as this seems to be our most frequent point of failure. Saying that the machine and network have failed multiple times, but the services have remained up throughout.

In addition to the servers, we also added mysqlrouter to the mix, in the initial design it's running on the same container as the mysql server. in hindsight, it would have been better to have a separate container for this, and the next phase the mysql servers will be hosted on seperate VPS's, and the mysqlrouter container will be running on the application server VPS's.

File Replication

We did some quite extensive testing of clustered file systems, including getting the application up and running on gluster. This again however proved to be a performance issue, and we found that gluster killed both CPU and memory usage. 

Eventually, we settled on a multi-pronged approach, the first being unison for two way synchronization. The second being splitting the file system into 'active areas' and archive areas. Our applications generally create files in directories based on YYYY/mm/dd - so a simple script was written to move directories older than a few days from the 'hot' storage area which was replicated using unison (based on inotify watches) and a 'cold' area, that was kept in sync daily using rsync. Softlinks were then created the hot file areas to point to the correct place in the cold storage.

This meant we could handle quite a bit of file activity as one of the applications is constantly creating files, and have those files available on multiple servers. For the next phase of development, we will be running unison in multiple containers for each pair of replication targets. And also considering NFS servers over TCP rather than replication for our main two front end servers.

Private Networking

One of the early issues before we set this all up was to work out how all these different servers would communicate, securely with each other. Normally for private networking, we had used OpenVPN. This is a client-server spoke system, however for a reliable network we would not want to have a single point of failure, and writing scripts to flip between different OpenVPN servers if something failed seemed rather messy.

To solve this we came across tinc, which solved our redundancy problem brilliantly. Tinc is a mesh VPN, which, in theory, can route around broken connections, so with servers A,B,C - if the line is down between C&A then it will route via B. It, as we found later does not handle a 'poor' (dropped packets) connection between C&A very well. You also have to make sure all the firewalls are correctly configured as if you incorrectly configure access to 'C&B', in that 'C&B' can see A, but A can connect directly to C&B, the network will work, however, will fall apart as soon as C goes down. It's a real, cross the t's and dot the I's network, get it correct otherwise when it fails you will be hunting down the issue for a while.

This is a map of the current configuration

Posted by in | Add / View Comments()

03 Jan 2019

GitLive - Branching - Merging

As things have slowed down in the new year, I've decided to give this blog a sparkle of life briefly. So if you are interested in engaging our services feel free to send us a message. As we have spare capacity at present.
Almost 9 years ago, I created a little application called gitlive, it's aim was to replicate our old subversion environment, where we mounted the subversion server over webdav, and whenever we saved files, they where automatically committed to the revision control system.
Posted by in | Add / View Comments()

28 Oct 2016

PDO_DataObject Released

Coding was complete last month, and has a huge test suite to covering a large proportion of the features. This should mean that replacing DB_DataObjects will be pretty easy.
You can either just checkout the code from github  / PDO_DataObject , or use the pear channel 
#pear channel-discover roojs.github.com/pear-channel
#pear install roojs/PDO_DataObject-0.0.1 


I revived my old PHP_CodeDoc code  (That needs publishing). It seemed simpler than trying to use any of the other tools out there. It's a pretty simple tool to extract structure, and documentation comments from the PHP source code. I added a small amount of code to export to our 'Roo UI bjs toolkit format' 
The generated files are pure JSON, and mostly contain the contents from the comments un-formatted. I decided that doing the Markdown conversion in JavaScript was far simpler (I refactored https://github.com/chjj/marked slightly for use with our libraries)
There are a few other tweaks I made, using `@category` to group the documentation, and writing category pages (using roobuilder), then putting it all together the index.js file loads the parts, and renders the manual.
This week I finished tidying up the rendering on mobile, and making sure all the comments render nicely using markdown. The result should be a nice easy to read and use manual.

Posted by in PHP | Add / View Comments()

17 Aug 2016

PDO_DataObject is under way

Work has started on revamping my PEAR package DB_DataObject, While it's served well over the years, and I still use it every day.. We have been funded to create a new version, which runs on PDO.

There is a Migration plan in the github repo for PDO_DataObject, I have currently completed the first two blocks, and almost the third block. But the key features are
  • General Compatibility to DB_DataObject with a few exceptions -  methods relating to PEAR::DB have been removed, and replaced with PDO calls
  • New simpler configuration methods, with some error checking
  • A complete test suite - which we will apply to DB_DataObject to ensure compatibility
  • Chaining for most methods so this works
$data = PDO_DataObject::Factory('mytable') ->autoJoin() ->where("somevalue not like 'fred%'") ->limit(100) ->fetchAll();
  • Exceptions by default (PEAR is an optional dependency - not required)
  • It should be FAST!!! - standard operations should require ZERO other classes - so no loading up a complex set of support classes.  (odd or exotic features will be moved to secondary classes)
Feel free to watch the repo (we are using auto commit, so the commits are pretty meaningless at present)

Posted by in PHP | Add / View Comments()

19 Nov 2015

Mass email Marketing and anti-spam - some of the how-to..

I'm sure I've mentioned on this blog (probably a few years ago), that we spent about a year developing a very good anti-spam tool. The basis of which was using a huge number of mysql stored procedures to process email as it is accepted and forwarded using an exim mail server.

The tricks that it uses are numerous, although generally come from best practices these days.

The whole process starts off with creating a database with

  • 'known' servers it has talked to before 
  • 'known' domains it has dealt with before.
  • 'known' email address it has dealt with before.

If an email / server / domain combo is new and not seen before, then apart from greylisting, and delaying the socket connections we also have a optional manual approval process. (using the web client).

Moving on from that we have a number of other tricks, usually involving detecting url links in the email and seeing if any of the email messages that have been greylisted (with different 'from') are also using that url.

On top of this, is a Web User interface to manage the flow and approvals of email. You can see what is in the greylist queue, set up different accounts for different levels of protection (either post-delivery approval, or pre-delivery approval etc..)

This whole system is very effective, when set up correctly. It can produce zero false negatives, and after learning for a while, is pretty transparent to the operations of a company. (email me if you want to get a quote for it, it's not that expensive...)

So after having created the best of breed anti-spam system, in typical fashion, we get asked to solve the other end.. getting large amounts emails delivered to mailing lists.

If you are looking for help with your mass email marketing systems, don't hesitate to contact us sales@roojs.com

Read on to find out how we send out far to many emails (legally and efficiently)

16 Nov 2015

Hydra - Recruitment done right

For the last few months we have been finishing up the first round of work on the Hydra Jobs platform. Something, along with the founders we think is quite revolutionary idea in IT recruitment. 

Key to it's design is the idea that the first step in finding someone is not putting up an advert, and expecting a shitstorm of resume's that are totally unconnected to the requirements. Taking a step back and realizing that as an employer, you would rather do a search for all the available candidates, than risk the time and wasted effort in sorting though unrelated piles of CV's.

We have spent the last 9 months working to get this to a MVP. The platform is now running, and the business operations are now underway.

So to make this work, the first step on Hydra was to design a set of Questions that could enable a detail search to work. What we ended up with is probably the easiest, yet comprehensive way of entering your profile data so it can be matched efficiently with companies recruiting staff.

It has been an interesting few months getting Hydra up and going, now we are over the hump of the work, we are looking for more interesting projects to take on, so if you know of any, please contact us.

Read on for some of the tricks we used to make this project, one of the best recruitment platforms around.

20 May 2015

More on syntax checking vala - and a nice video

As I wrote last week. I had added full syntax checking to the editor. So it runs a full compile check as you type.
Here's a nice video of it working...

After the initial joy of adding this to code, I soon realized it had a fatal flaw, read on to find out more..

Posted by in Gtk | Add / View Comments()

09 May 2015

Fetching Resources from github in the App Builder and fake web servers

My final words this week on the builder - handling resources, and fake web servers

While I talked in the other posts about how the builder extracts the API for various components from the libvala library and the vapi files, some information that the builder requires has to be manually, created or fetched from other locations.

When the Builder was written in seed, it basically looked at the source code directory, and read files relative to the source code. For the Vala version however, it's not expected to know about the source code directory, so I had to use a different approach.

Posted by in Gtk | Add / View Comments()

08 May 2015

libvala testing code and extracting API from the vapis

And the next part in the series. Gir and Vala structures, Nothing like a slow day to write a few blog posts. 

The App Builder was originally designed to build applications using seed (the gobject introspection webkit javascript engines bindings), One of the key elements of how this was done involved introspecting the Gtk API, and extracting all the properties, signals and class structure.

In this post I will go through the history of how I extracted the API information on Gtk, initially from Gobject introspection and GIR files, upto the current version which uses libvala to get the correct API direct from the vapi files.

Posted by in Gtk | Add / View Comments()

07 May 2015

App Builder - Database based Plugin builders for Web components.

It's been a busy month, unfortunately not for our paid work, which has dropped down to a trickle. Taking advantage of this I've been building more into our App Builder. This post hopefully is the first in a series about some of those additions.
The Primary purposes of our Builder is
  • A WYSIWYG tool for web applications using both Bootstrap or the RooJS libraries.
  • A new visual way of building Gnome/Gtk Applications 
In working towards these goals the builder has moved forward in a few directions. the first one that this blog post talks about is generating User interfaces from Database Schemas.

Posted by in Gtk | Add / View Comments()
   (Page 1 of 24, totalling 231 entries)    next page »

Follow us on