AlienCarnivore

Why does this website go down so much?

9 posts in this topic

I don't want to sound ungrateful for all the hard work people do for this website to keep it up and running. The wiki is invaluable to me and the forums here are very informative and friendly. But I've noticed this website goes down quite a bit. I try to connect to it sometimes and a page telling me the server could not be reached appears. Sometimes the site is down for hours at a time. Is there a particular reason for this? I don't have this trouble with any other website I go to.

Share this post


Link to post
Share on other sites

There could be a few reasons, but here are the most common ones:

* Bandwidth: Accessing a website causes you to download information, which also must be transmitted from the server. Some servers, such as those owned by Google and huge companies, have their own dedicated data lines. Without such a connection, peak times of Internet usage will affect the server of this website. If lots of people in the same neighbourhood as the server are using Netflix on a Sunday night, that slows the server's speed down.

 

* Denial of service: Only so many people can connect to the same server at the same time. This is determined by the level of data within the server. This website probably has a limit much smaller than a very large website and its associated infrastructure, like Wikipedia.

 

* Cannot ping the server: Browsers will ping a website before loading it, just to make sure that the URL is legitimate (instead of infinitely loading the page that might have a spelling error and not doing a thing). A super-dooper expensive server will have more than one network card, so if one misses the "ping" the others will get it and respond. Otherwise, your browser assumes the page doesn't exist.

 

Finally, the last thing is related to the other three:

 

* Cached files: Your browser may have saved a cached version of the website that is from when it was down. So unless you force a reload or dump the cache, it will just load the error screen instead of loading the page afresh.

Share this post


Link to post
Share on other sites

Since traffic is much lower when the show isn't on, we reduce the number of servers for the website during this period. This means we can sometimes run into issues with how much the site can handle. Additionally, we have some configuration issues related to memory right now that we have not been able to solve and that has been the cause of the recent crashes.

Share this post


Link to post
Share on other sites

As a follow up, after another inexplicable crash of our caching system (memcached) leading to a long downtime over the evening, we have gone ahead and installed Redis and set that up to see if it will be more stable. We have not moved all parts of the site (the wiki and Westeros.org-proper) to Redis yet, but that is currently the plan. In the interim, we're hoping that loading memcached with less work will stabilize it and we should be more stable overall going forward.

Share this post


Link to post
Share on other sites
On ‎2016‎/‎11‎/‎20 at 8:31 PM, Linda said:

Since traffic is much lower when the show isn't on, we reduce the number of servers for the website during this period. This means we can sometimes run into issues with how much the site can handle. Additionally, we have some configuration issues related to memory right now that we have not been able to solve and that has been the cause of the recent crashes.

On a related note, is it just because of the absence of the Show, or is traffic down in general as Winds of Winter drags out longer and longer? It certainly seems to me that new topics in the Ice and Fire section are growing fewer, and that the regular, long established members are dropping in less often. Other than Lord Varys, most of the long term debaters seem to have largely vanished.

 

Share this post


Link to post
Share on other sites
2 hours ago, Free Northman Reborn said:

On a related note, is it just because of the absence of the Show, or is traffic down in general as Winds of Winter drags out longer and longer? It certainly seems to me that new topics in the Ice and Fire section are growing fewer, and that the regular, long established members are dropping in less often. Other than Lord Varys, most of the long term debaters seem to have largely vanished.

 

I have noticed this as well, very few of the guys I saw most of from when I was most active in mid-late 2015, personally I mostly just lurk now, but I don't really know most of the "faces" anymore.

Share this post


Link to post
Share on other sites
16 hours ago, Ser Arthur Hightower said:

I have noticed this as well, very few of the guys I saw most of from when I was most active in mid-late 2015, personally I mostly just lurk now, but I don't really know most of the "faces" anymore.

Yeah, same here. I've always been a lurker more than a poster. Generally if it's report-writing time (since I'm a teacher) then I post here as part of my procrastination.

Once tWoW is out, though, I'm going to be here 24/7.

Share this post


Link to post
Share on other sites

It is quieter here, and from what I can see it is quieter at just about every other ASoIaF site than at a similar time last year. Lots of possible reasons, but certainly the fact that there's not been any new ASoIaF material in a year -- and no new novel in the series for more than that -- certainly has something to do with it. We expect it'll zoom right up once TWoW is announced as completed, however.

Share this post


Link to post
Share on other sites

Further update as we have had two down times the last few days.

 

The first one was caused by the Redis memory management system -- the new one we installed -- running out of memory. It's a new system for us and we needed to do some tweaks to prevent it from eating all the available memory. We've now done those tweaks.

The one yesterday, on the other hand, is less explicable now that we have its memory properly limited -- it just seems like Redis crashed for no obvious reason. There are ways to get detailed debug logs but we'll have to wait a bit for that. It seems to be behavior similar to what happened to memcached as well, so ... something's going on, perhaps with the forum software throwing out garbage that the cache system can't really parse. We'll see. Until then we plan to periodically restart Redis just to see if that helps the situation any.

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.