You may be looking for the Nation of Earth, the Ghostly World, or possibly some other sites that no longer exist - and you have ended up here. Here's a bit more info on why that is the case.
My server (affectionately named Lillith) froze up yesterday while handling various website requests, resulting in some undue panicking on my part as I tried to imagine having to rebuild everything from scratch. Even though I keep daily backups and such, I'm terrible at documenting everything I set up on the server, consequently it's about 3 years of random configurations, installations, etc. that would probably take me a few weeks to replicate. I mean - if the server should suddenly die I would probably be able to get the websites back up within a day or so, but other functionality such as email and so forth would be a bit more painful.
But as it turns out the server was fine. MySQL had recently been upgraded when I was doing security updates over the weekend, however, so I personally believe that this upgrade was the culprit (or more accurately, my failure to properly configure the upgraded version was the culprit). This was not an immediate conclusion that I reached however.
My server is in an isolated part of our home and does not have a monitor, keyboard or any other connections aside from a power cord and network cable. I access it remotely via SSH and pretty much handle all changes that way. Starting at about 9:30am yesterday it was impossible to open a web page, and I could not get SSH to complete a connection. I also have VNC configured for remote access to my desktop computer so I quickly realized it was not an issue with my internet provider when I was able to sign into that system. I was concerned that the server might be experiencing a DoS (denial of service) attack of some kind so I cut all access to it at my firewall, and attempted to sign in via SSH from my desktop computer without success.
I tried calling the house but we screen our calls and someone (one of the cats perhaps?) had accidentally disabled the answer machine and Brig was busy tending to our youngest so I wasn't able to get through. I then decided that I would rather leave the server running in its current state so I could hook up a monitor and keyboard to it when I got home and get a better idea of what was going on. Of course, life kicked in at that point and after a trip to our CPA to get our yearly taxes squared followed by dinner at our favorite Mexican restaurant, El Azteca, we finally got home and I was able to take a look at the server at around 9:30pm.
Upon getting the monitor and keyboard attached I immediately saw low memory errors as Apache was trying to kill tasks in order to process others. The hard drive light was solid red and the drive was very noisy as it was obviously chewing through swap in an attempt to mitigate the low memory problem. It took three minutes to reach a login prompt, and then I waited another five minutes to get to a command prompt with the idea that I could stop Apache from running and figure out the issue from there, but I finally gave up and reset the server.
While I was waiting I looked up the error messages I was seeing on my Droid and came across a Drupal comment that seemed to tie into what I was experiencing, and was also dated March 4th, which was close enough to the date of when I performed the MySQL upgrade that it seemed likely to tie into my problems. I'm not using a virtual server, but this information still seems to apply to my configuration.
When I first hooked up the monitor I was seeing error messages similar to this one:
Killed process 18762 (apache2) vsz:187372kB, anon-rss:20172kB, file-rss:380kB
Out of memory: kill process 18740 (apache2) score 46907 or a child
Not an exact error but I had searched for "killed process apache2 out memory" and found several articles, the one linked above being the most pertinent. So as per the comment author's suggestions I followed these steps below.
I first edited /etc/mysql/my.cnf to include the following settings:
key_buffer = 16K
max_allowed_packet = 1M
thread_stack = 64K
table_cache = 4
sort_buffer = 64K
net_buffer_length = 2K (I had to add this line as it was not already there)
The article also suggested adding the line "skip-innodb" to my.cnf if you don't use InnoDB tables. I am using Roundcube webmail, however, which uses InnoDB tables, consequently I did not add this line.
I next ran this command to find out what type of MPM is used in my Apache so I could edit the correct section of the config file:
apache2 -V | grep 'MPM'
My installation uses "prefork" so I edited that section of /etc/apache2/apache2.conf with the following:
StartServers 1
MinSpareServers 3
MaxSpareServers 6
ServerLimit 24
MaxClients 24
MaxRequestsPerChild 3000
I then restarted both apache2 and mysql with the following commands:
sudo /etc/init.d/apache2 restart
sudo service mysql restart
The comment also suggests updating your Spamassassin configuration if you have one - but mine is currently inactive. I also need to reduce the size of my swap partition to something a bit more manageable - it's currently a bit too large for a typical server.
However, at this point everything seems stable, and has continued been so since last night. But while I checking things out I decided it was time to tidy some other things up.
My friend fil had asked me to take his website, the Nation of Earth, offline. I had been hosting it for him as part of my Drupal installation, and while it gathered some interesting people and resulted in some cool discussions, it was also a maintenance headache as the site was continuously attacked by spammers trying to insert dangerous hyperlinks into comments across the site. Even with troll, bad behavior and recaptcha modules installed, along with restrictive comment support and limited access provided to new user accounts, the site continued to require periodic maintenance to remove unwanted content and block offending users and IP addresses. This sort of activity is like trying to prevent the tide from destroying your sand castle - it never ends.
So as per his request I exported his user contacts and nodes into readable format and sent him copies. I also backed up the site files and SQL database and sent him copies, then removed them.
My other site, The Ghostly World, had been registered to ghostlyworld.com but I had opted to not renew the domain name registration last year due to lack of free time to work on the project. I had migrated the site to ghostly.fatedtoend.com and had promptly forgotten about it. Based on this I decided to pull the trigger and take it completely offline in the same fashion.
There were some other lingering SQL databases that were still getting backed up nightly even though they were not tied to any sites, so I archived those and dropped them. I am now only managing three Drupal sites including this one - the other two are my wife's blog Dakini Hikwsi and my MUD site, Meat MUD Classic - which is soon going to get similar treatment as it is replaced by the new site for Havens MUD (when I get around to it). I also have registered my last name as a domain and plan to build a personal portfolio/resume site around it in Joomla (which is probably overkill for such a project but I've been meaning to check it out and get more familiar with that CMS - it's about as popular as Drupal for a reason).
So there we are - virtual spring cleaning on my server. The size of my total Drupal installation is now much smaller, which means less needs to get backed up nightly. This will also mean my backups don't need to be purged as frequently. In addition, less sites are active, so there is less load on the server, which means everything should be hunky dorey (for now). Heh (crossing my fingers).