This server is a ntp server located in Cleveland, Ohio, US and is part of the ntp pool project. You can always use this server by using the name ntp.vrillusions.com. You can also use one of the ntp pool addresses mentioned below to randomly use a participating time server.
This server is an active member of the ntp pool list of servers. It is listed for pool.ntp.org, north-america.pool.ntp.org, and us.pool.ntp.org. Additional statistics are kept by the ntp.org site as well.
Below are some graphs of the current time offset and estimated number of active clients (can only be estimated since ntp protocol uses udp). You can use this server as your computer's time server using the hostname ntp.vrillusions.com. You can see whenever the server is included in the dns for pool.ntp.org or us.pool.ntp.org by the sudden spikes in usage. I've also included the list of upstream servers for reference. The one with the * is the current reference server. Graphs are updated every 5 mins.
time offset

packet counts

remote refid st t when poll reach delay offset jitter ============================================================================== *209.51.161.238 .CDMA. 1 u 226 512 377 16.889 0.261 0.614 -64.236.96.53 .ACTS. 1 u 22 512 377 10.579 2.089 0.148 -18.145.0.30 .PSC. 1 u 211 512 377 18.915 0.493 0.124 -67.128.71.76 .CDMA. 1 u 491 512 377 29.263 4.359 21.151 +18.26.4.105 .CDMA. 1 u 172 512 377 18.975 -0.175 1.027 +132.163.4.103 .ACTS. 1 u 507 512 377 35.360 0.002 0.698 -132.246.168.3 .PPS. 1 u 156 512 377 30.602 2.298 0.260system uptime
11:41:02 up 107 days, 13:30, 1 user, load average: 0.25, 0.29, 0.26
FAQ
- Why did you setup an NTP server
- It was mainly just because I wanted to. Also the load this places on a server is minimal and has the added benefit that I can use this server for time sychronization between all the computers at where I work.
- How do you generate all the graphs
- The actual graphs are plugins for munin that are copied from the admin area to the public folder. The data for offset is from the command `ntpq -c rv` and packet counts from `ntpdc -c iostats`. I haven't bothered with creating a separate zip file, but you can get the scripts to generate the graphs using subversion. In linux you can type `svn co https://svn.vrillusions.com/svn/munin-plugins/trunk/plugins` (without the `s, and just say permit on the SSL certificate since it's issued by CaCert which isn't an "official" CA).
- How long have you been a member of the pool.dns.org service
- Since mid September, 2006. A replacement server was installed in mid March, 2007 with a new ip address.
- What are some stats on the server used
- This server is a dual P4 1.266 ghz with 1gb ram running Ubuntu Server with a 10mbit+ internet connection. NTPD is version 4.2.4p4.
- Any other sites with statistics
- Another really nice site with ntp statistics is lx.ujf.cas.cz--it was the inspiration for the graphs I did and how to go about collecting the data. Also the NTP Pool project's How to join page lists several servers with statistics.
- Ever thought of being a stratum 1 server.
- While a stratum 1 server (a server that obtains it's time directly from some method) is tempting, the cost for the hardware is very expensive. Also the datacenter the computer is located at probably couldn't get any of the radio signals and they're not going to let me drill a hole in the ceiling to run an antenna to the roof. I have looked into it and a CDMA time reference is just too expensive for me. GPS reference systems are less expensive at < US$100, but GPS needs line of sight which won't work in the datacenter. CDMA can go through walls and I get a great signal inside, so I know it would work, but it's a lot more expensive than GPS.
Status
- Sep 1, 2008 - changed maxpoll back to 512 seconds. Seems to be fluctuating a lot lately.
- Aug 20, 2008 - server is now IPv6 capable. This won't work from ntp pool (only has the IPv4 listed there) but if you add ntp.vrillusions.com and your server queries for IPv6 addresses it should find it.
- May 23, 2008 - upgraded server from Ubuntu Server 6.06 LTS to Ubuntu Server 8.04 LTS. NTP was down for a couple hours but it's back up and running along just fine.
- Mar 25, 2008 - removed the prefer tag I had on one of the servers. Will see how much the time fluctuates when ntpd can choose whatever it wants.
- Mar 20, 2008 - Created a convenience subdomain for those that wish to use this server directly.
- Mar 4, 2008 - Stopped collecting the ntp client stats. It's not as amusing with the new system and it puts a lot of load on the server. Packet counts are more accurate since they are reported by ntpd itself. The last image of ntp clients can be seen here. The client stats scripts I did use are available here.
- Jan 1, 2008 - Removed the minpoll 9 option from servers so it defaults back to 1024s and see if it still fluctuates as much.
- Dec 23, 2007 - Added system uptime stat.
- Dec 19, 2007 - Removed all the servers I was checking for reliability. Added one new server and set the prefer tag on one of the servers to help eliminate jumps between servers. Now the server that I was going to replace because of unreliability is stable again. So guess I just have more candidates.
- Dec 11, 2007 - Added a bunch of servers that I'm going to monitor for reliability.
- Oct 8, 2007 - Took server down to move racks accross the datacenter. Back up now, with the usual initial spike on startup.
- Sep 27, 2007 - Since there's a new way servers are put into the pool based off of their netspeed, I've upped mine to 100mbit. At previous 1.5mbit there were very few spikes in traffic.
- Sep 19, 2007 - Noticed within the past few days the upstream servers have been varying a lot, hence why my offset has been more sporadic. I've started to look into other upstream servers that should stabalize things.
- Aug 10, 2007 - Took machine down to just install the secondary power supply and mount the second hard drive properly. Was supposed to be quick but had some issues getting the network back up. Then aparently the ntp service didn't start at boot and didn't check till a few hours later. Everything is back up and running now.
- Jun 3, 2007 - Noticed that yesterday around 11:15 AM EST the server had a 49.6k clients / 30k packets/sec spike. I increased the memory limit a week ago and looks like the new memory limit is good (32mb)
- Apr 20, 2007 - Did some more upstream server juggling. Turns out one server that has a great stable response most of the day will start failing mid afternoons. This would cause the offset to start going out of wack. Plus two other servers were consistently off by large amounts. Tried a few new ones last night and see how things go.
- Apr 19, 2007 - Today there was a 22.5k clients / 208 packets/sec spike and didn't get any out-of-memory errors, so looks like it is fixed.
- Apr 14, 2007 - NTP crashed, ran out of memory. Before crashing it was serving 28k clients and 210 packets/sec. Surprised because as far as I can tell there is no memory limit set. Manually set it and see if that fixes it.