Server Update

[Hells Warrrior]

Today we purchased a new package with our current hosting provider and we are moving to a new server that will be running on CentOS7 64 bit operating system. We will be doing some optimising / upgrading software and basically getting things running as best and as fast as we possibly can. Moving to this new setup will bring us better performance and greater scope for optimising.

Recently we have updated our database to run on MariaDB, we have also moved all the tables for the database to InnoDB, we are also utilising cloudfare for static content.

What does this mean?

What it effectively means is we will need to update our nameserver to the new IP address, it means we will experience a little bit of downtime whilst everything is migrated and DNS propagation occurs (usually 24/48hrs) - some won't experience any downtime, others may find access is lost for a period.

This a necessary step in getting the site performing to the best of it's ability and making things as fast as possible for all our members and staff. I'm sure you'll all survive without access but will enjoy the benefits it brings. When we do migrate we'll take the site offline and once it has migrated it will remain offline for a few hours until we tweak things and get the best response times we can possibly achieve.

We thank you in advance for the understanding, we'll be looking to start this process tomorrow/Saturday and we'll update you all in due course.

[Hells Warrrior]

We'll be putting the server in lockdown in a few hours time, approximately 10pm GMT to start the migration process, it shouldn't take long but we are not wanting to have any data added during the migration process which could ultimately be lost as a result of migrating.

Expect us to be offline for a minimum of 24hrs - you will be able to access the site but presented with the offline message.

[Lenne]

Can't you do this during the week? 

Just kidding, friend. :P

[Hells Warrrior]

I can't I could do it during next week but in all honesty, we are now paying for two servers and we need to make the change ASAP. I appreciate it's when you return every weekend but that's how the cookie crumbles my friend :p

[Hells Warrrior]

Due to plans not going as expected, we have enabled the site on the current hosting package again. We have not been able to have the nameservers updated at the registrars to point to the new server and we are not able to retain the existing IP addresses assigned to this package to the new package.

As a result, we have migrated the site over onto the new package and all that needs to be done when the time is right, is to transfer the database when the nameservers have been updated.

We apologise for this, however, we will be doing this again shortly and we thank you for the understanding and patience.

Posting abilities have been enabled again.

[Hells Warrrior]

I apologise for the extended downtime today, the DNS settings all got amended today, which we requested but I was unaware had been actioned til I got home. Then they needed to propagate throughout the internet and is still occurring,

Then unfortunately it came to light that the SQL users had not been transferred as part of the migration which we had to correct, then we had an issue with our memcache settings which speed up the service to the people visiting. So we have corrected these but we also need to add the correct style to the site (which I have saved on the PC at work).

Please bare with us / me whilst I implement all of this but the benefits will start to show soon. 

Thanks for the understanding.

[NaBrZHunter]

The site looks badass, man. GREAT WORK!

[Hells Warrrior]

I think I'm going to disable background when viewing the site on a mobile (not mobile desktop, which will return later today and be updated).

I got to 2am and my eyes where being held opened by match sticks.

[PINNAZ]

Good job @Hells Warrrior. Nice background & header image too.

Frustrating that it didn't go to plan for you. I appreciate the time you spend updating the site.

 

(P.S. - Why is the shit editor back? The undo format & spoiler buttons are missing)

[Hells Warrrior]

Should be back to normal now with the editor bud, that shouldn't have changed to be honest but it did and I've changed it back.

[Hells Warrrior]

So for those that have been wondering, what is it that is actually happening, why the change, why the downtime and why the urge to get better performance?

The main reason is TTFB (Time To First Byte) and page load (full) - the principle behind this, is the faster a site loads, the better the overall experience to the end user (you and me). So this is the snapshot of the move to the new server, indicating page load.

What are we looking at?

In the snapshot above, you will ratings in the top left, A, for example is good, C is average and X is not not being used at all. We are scoring well for everything apart from caching static content (which we are using, but only Zend Op Cache, we will be utilising Memcache again and we will be using CDN (Content Delivery Network via cloudfare) again.

However, this is about bench marking without using the additional items. 

As you can tell, the first byte is now looking good, I want it to be better and that requires additional optimisation in different settings, specifically optimising how the database performs and utilises the memory available correctly.

We need to reduce the page size on initial load, as it's loading at 1MB and I want to reduce this by 50% (if possible and without affecting the overall look). I want to reduce the amount of requests on the initial load as well, currently it's sitting at roughly 71, I want to get that down to between 30/40 request, the less requests the better.

What we also want to do is make the site snappier, i.e. click a link and within a second the page load happens and you can start reading the next page/topic. Our limitation is the server and the amount of processing cores available and memory available. As we are on a VPS (Virtual Private Server) we are basically on shared hosting / the big difference is a VPS is effectively a dedicated server within a shared hosting enviroment.

What we want and it will happen in the future, is a dedicated server - where all the processing power and memory is available to us and us alone. However, dedicated servers cost a bit more money that VPS (roughly $100 more expensive per month).

However, we are heading in the right direction and moving the site to the new VPS package running on a 64 bit OS which allows us to better utilise waht we have, has been as step in the right direction. Over the coming days and weeks, I will be doing more optimising but this won't affect site uptime (unless I need to borrow the lead for the kettle).

[Hells Warrrior]

Apologies for the downtime yesterday, we experienced a hardware node failure and this caused the error seen yesterday. We had to do a recovery on SQL as it failed to restart.

All should be good now, thanks for the patience and understanding.