As many of you have noticed, as of late ShoutWiki has been displaying more than a few HTTP 503 errors (“Service Unavailable”.)
This blog is going to outline what these issues are and why they are caused, what we are doing and why this blog has been published now. Let’s start with the latter. The reason we’ve left it so long is because we had originally planned to be able to give you all the facts on, in particular a timeline related to the solution. Unfortunately we are unable to give you these dates as of present – although as soon as we are aware we will let you know – and we feel it is wrong to keep our users (you) in the dark for even longer.
To keep it simple, the issues are being thrown by our Varnish caching server as the Apache server (which servers you the pages and images) has been unavailable. In short, this has been caused because the web server has been using excessive amounts of CPU and RAM, and the whole server has been left without memory.
What have we done so far? We’ve temporarily disabled various high CPU processes (unfortunately this includes export functions) in an effort to preserve and extend the uptime of the site. In addition we’ve taken steps to make the infrastructure more scalable. We’ve been making steps towards making it easier for us to add additional web servers to our setup by moving non-wiki sites to their own server, and we’re looking into moving images to the cloud. Obviously this takes a fair amount of planning due to the financial and technical implications.
And finally, what we’re going to do. Simply put, we’re going to reinstall the operating system on the server. We’re fairly certain that there is a problem with the template the web server VPS uses. While we do this, we’re also going to be switching from Apache to Nginx in an attempt to improve performance. When is this downtime, and how long it will it take? We’re not entirely sure – which is what we’ve been waiting on. Our web server VPS is attached to the account of a former staff member, mainly for financial reasons, but also due to the connections this former staff member has with our provider. Unfortunately, we’ve been unable to get a hold of them recently (as soon as they are around again, we’ll be able to get the apache server back online!), and we are still waiting on a reply on when he is available to reinstall the operating system. The length of downtime will be determined on how much data has to be shuffled around, currently images take up a large proportion of the disk space, which is a big reason in why we want to relocate our images.
ShoutWiki is a for profit organisation, however none of our staff members get paid for their work, they are all volunteers and all our income is reinvested into servers and this will be the case until we can establish a fast and stable service.
Again, we apologise for the downtime, and thank you for your continued patience. We understand this is a very frustrating time and we look forward to resolving the issues as soon as possible and allow you to get back to editing your wikis.
On behalf of ShoutWiki staff,
— Lewis Cawte
Chief Technical Officer, ShoutWiki.