WP Super Cache not working in Apache 2.4 with load balance

The plugin won’t work well in a multi server environment as it writes files to the filesystem. That’s why each server has a different timestamp as the files are stored on each. You could try writing the cached files to a shared NFS directory. There aren’t any locking issues any more but you’ll have a single point of failure again.

I’m not sure why you’re getting 500 errors though.

I am using a VPS, and was thinking up upgrading from Apache 2.2.25 to Apache 2.4.6

Are you saying I should not do this?

I am using a VPS, and was thinking up upgrading from Apache 2.2.25 to Apache 2.4.6

Are you saying I should not do this?

No, I don’t think he meant that. He just expected those timestamp related errors for this use-case, because…

The plugin won’t work well in a multi server environment as it writes files to the filesystem. That’s why each server has a different timestamp as the files are stored on each.

I wouldn’t recommend WPSC on a load balancing environment, either.

Thanks.

After chatting with my hosting company, I tried several times today to upgrade my server to 2.4.6, but it kept failing. Not sure why. So then they did it. Now it seems to be working.

This is what I expected to learn and it raises a major question.

Godaddy is the world’s largest hosting company. They are now calling their single port server environment legacy. They are migrating everyone over to their new grid load balanced system. Any new customers joining GoDaddy will automatically be put into the load balanced system. If Godaddy is doing this, then this is the direction everyone will be moving to.

So if single port servers are legacy, that means WP Super Cache is now officially a legacy plugin.

Are the developers of WP Super Cache going to write the code so their program works in the new normal load balanced environment? Yes, the code can be written to cache with matching time stamps in this environment. Someone just has to be willing to write and test that code. This isn’t really new. It’s just new to the hosting companies. A company I worked for installed a fully load balanced system in 2004 for its server environment.

What other caching plugins are currently written to work in a load balanced environment?

As Bob Dylan sang in 1963, “The times they are a changin.”

What other caching plugins are currently written to work in a load balanced environment?

Batcache is the most popular. WP FFPC is used in one of the Juju charms for WordPress.

Theoretically, if you set up WP-FFPC with memcached backend, all hosts connecting to the same memcached pool, it should work, regardless of the number of servers, but this is theory.

If you could prove it, it’d be very happy ??

p.s. I’m the author of WP-FFPC

donshapiro – do you have any documents or press releases giving more details? I find it hard to believe they’d move all their customers to load balanced servers because of the complexity involved. A basic hosting package certainly doesn’t need to run on several load balanced servers.

Even if that were the case there are still many websites not on Godaddy so it’s premature to say WP Super Cache is a legacy plugin.

Donncha,

You’ve asked a very good question. There is a critical reason all websites should be in a load balanced environment…port access to their website. Under a legacy system, all traffic to a website must go through one port. If that port gets busy either because the website (hopefully) is growing or because the website is on a shared server, then higher traffic levels will cause the sites to slow down. Sometimes, those slowdowns can be dramatic.

Even those who have paid more for a dedicated IP and hosting package are housed in a server with hundreds and maybe thousands of other websites. Even if someone has the money to run on their own server which very few WordPress websites do, they are still subject to speed issues through a single port.

Load balancing of packets and traffic is the best way to insure that a website of any size and volume doesn’t experience slow downs due to port traffic.

Why is this so super important? Because site speed is one of the key factors Google uses in page rank. They want to rank pages higher that are a better experience to the user. You know this which is why you created WP Super Cache. But caching can’t help speed of delivering websites if the port is clogged up.

When I called Godaddy and told them we were experiencing speed issues, they immediately asked us to migrate to the new servers. There was no cost. They are additionally encouraging us to consider migrating to CP Panel which they just started offering instead of their own hosted control center. They said that would increase our speed too.

Back in 2004 when I worked with a super server techie (who had designed American Express’s server center amongst other things), he said load balancing is the wave of the future. It has now arrived in the hosting environment and will become more popular both because the hosting companies want to move people to these grid environments and because website owners want to keep their speed high and avoid bottle necks.

I would be surprised if there are any non load balanced server environments for the medium and large hosting companies within 10 years, most likely sooner. No one is going to build a new server environment today that is not load balanced. It is only the old systems that are that way. As those are replaced like Godaddy is doing, they are being replaced with grid servers. As the majors move, so the minors will follow.

For the hosting company, this also gives them advantages in how they administer their network and the level of availability they can consistently deliver. One huge advantage for a hosting company is that a server can go down without it affecting website availability. That keeps their customers happy and attracts more business. In a legacy environment, if the server has a problem, those sites go down until its repaired. They have to use mirror systems for availability to reduce the down time but that is more expensive to operate than a load balanced environment.

Hope this helps,

Don

Ah, that’s clearer. Not every site needs to move to load balanced hosting. It adds a ton of complexity. I wouldn’t go calling websites hosted on a single server “legacy hosting”. Sure, there’s a single point of failure but it’s not what the customers of that hosting company paid for.

I thought you were quoting from some official document but this is just speculation. Sure, in ten years time things will likely be more virtual than they are now but I’m quite sure there’ll still be basic hosting with one server serving multiple websites.

Donncha O Caoimh you’re mistaken, shared environments will be legacy within a year.
Fortunately, chroot-on-steroids environments, like lxc containers ( see heroku ) are rising and just like disposable aws containers, they will take over soon.

The complexity is not bad if it’s automatized ( eg. staging box -> auto distribution ) and loadbalancing could solve a lot of issues with the boxes, including maintenance time, single point of failure, and so on. They could easily provide a lot more 9 uptime this way, and it’s be foolish not to do so. Shared hosting environments are about to be legacy in a very short time.