I just wanted to take a few minutes and update the awesome STH community on the state of the site. Every so often I like to respond to some of the common mailbag questions and give updates. I figured I would write a bit about those here. For those wondering, suggestions do often make it to the OKR for the site so please do keep them coming.
Forums moved to SSL – why not the main site?
One of the major features we have seen over the past few months is that the forums have moved over to SSL. Since then there have been a few updates. First we have tracked down (hopefully) all of the mixed/ insecure content over the past few weeks. We also updated the SSH key to SHA-2 256-bit. For those that are not aware, Google is depreciating the older 128-bit keys so it was a necessary update.
During this move there were a few hours where some browsers (e.g. Android Chrome) were unable to connect to the site. The problem was insufficient testing. I was making the change myself and saw it working on desktop browsers. I pushed the change to production and then realized some time later that the issue was in place. Mea culpa.
A question asked often is “why not have the main site as SSL too?” Absolutely a great question that I have been asked quite a bit recently. The biggest issue for this is simply the setup we are using. Currently the setup involves:
User request -> firewall -> server -> varnish -> nginx
Varnish does not support SSL. That is presenting a bit of an issue since the common fix seems to be adding nginx reverse proxy in front of varnish making the process:
User request -> firewall -> server -> nginx reverse proxy -> varnish -> nginx
Testing has been mediocre at best. The performance hit is not too bad, but it is slower. STH is an “older” site now with well over five years of content. That content is not always HTTPS friendly so it is certainly a longer process than expected. The change is coming. Luckily, STH has virtually no user account so it has been a lower priority.
Where are the processor reviews?
For the Intel Xeon E5-2600 V3 series launch I did a series on Tom’s Hardware. I do have a backlog of CPUs to test. One of the biggest questions right now is how to present the Linux-Bench results. There are currently 11 tests in Linux-Bench and over 55 different data points. Running the benchmark is extremely easy. Parsing the data is the easiest it has ever been. The big question now is what to do with all of the data it generates.
Ideally there would be a set of results that could be easily filtered/ graphed by users. Another set of summarized results should be presented on the main site. We are still working on how to satisfy both needs. Any thoughts/ suggestions can be added here.
There was quite a bit of feedback that the previous STHbench.sh iteration created eye-charts. We are taking that feedback and hopefully creating something better.
For reference, the archive of STHbench.sh results can be found at http://linuxbench.servethehome.com
STH passed 1.5 million spam messages stopped by Akisment and averaging over 100K each month at this point. We are looking at ways to reduce forum spam significantly.
I certainly would say buying the subscription is highly recommended.
ServeThe.Biz – what is up with that?
About a year ago we started toying around with the idea of launching a ServeThe.Biz site. Some folks have seen the ServeThe.Biz site and wondered, what is going on there.
- According to demographic stats we have, 85% of STH visitors access the site from work
- More than 75% of visitors work in IT, programming or a similar technology field
- Most of the site’s content is really more of the higher end of prosumer and into business/ enterprise topics
- The #1 question I get from companies and other members of the press is “why home?”
There is a decent case to be made to change the site’s name. That involves a lot of heartburn. It also involves a lot of work rebranding. On an annual basis it is reviewed and rejected. Last year’s cycle yielded launching STB as a second site. One major issue occurred: lack of time and focus for both sites independently.
This year’s review is coming up again. More to come.
The Great Crash
Some readers may have remembered the great crash of 2014. The TL;DR of the issue is that while we were working on updating a few of the nodes, we were running the production sites from multiple servers within a Dell C6100 chassis. That chassis had a massive storage failure where Kingston E100 800GB drives failed in three different nodes. The best bet is that it was a backplane issue since the failure was cross node. That chassis is gone as are the drives.
Things are running again. There are now at minimum copies of the site on two drives per node, multiple nodes, no striped raid types (only RAID 1), and back ups, all on different types of drives. One thing we learned was that the Kingston drives died yet the Intel drives in the same nodes were fine.
Still not at the level of redundancy we want, but certainly a step in the right direction. On the plus side, this current setup has outlasted a memory module failure, a domain controller failure and a disk failure with under 30 seconds of downtime last quarter. Still not where it needs to be, but it is certainly heading in the correct direction.
Wrapping up this STH update
That is is for the October 2014 STH update. If you are a long time reader but have yet to get involved in the community head on over to the forums and join us for great discussion.