aprs.fi suffered a partial network outage last night (July 5 21.37 UTC - July 6 5.03 UTC). A gigabit Internet uplink was lost for the duration of maintenance by an ISP and the alternative path didn't quite provide all of the Internet's routes over BGP. Investigation continues.
The outage caused trouble for many to view the site, and some APRS data collection problems. The stats page gives a hint of the scale of the problem: http://aprs.fi/stats/daily
About 50% of the viewers disappeared. It would seem like regular APRS-IS data was collected just fine, but some CWOP data was lost. There are a lot of alternative routes that the data collector can take to get to the APRS-IS, and that seemed to work.
Sorry for the trouble.
The news of https://aprs.fi/ - new features and interesting attractions found in the APRS and AIS worlds.
Wednesday, July 6, 2011
Wednesday, June 22, 2011
Map of the APRS-IS
Have you ever wondered what the APRS-IS looks like? Well, here you go (click on the link, 1.2 megabytes, 3154 x 3072 PNG image):
Map of the APRS-IS
Map of the APRS-IS
- Servers with less than 5 clients are not shown.
- Only servers running javAPRSSrvr are shown.
- Darker colour indicates more clients connected.
- Green servers are core servers, orange ones are T2 hubs, red ones have a T2* name (most of them belong to the Tier2 network), purple ones are WXQA servers, the rest of them are blue.
- Inter-server links between hubs and core servers are blue, the rest of the links are green.
Sunday, June 5, 2011
Wed 8 June: World IPv6 Day
On 8 June, 2011, Google, Facebook, Yahoo!, Akamai and Limelight Networks will be amongst some of the major organisations that will offer their content over IPv6 for a 24-hour “test flight”. The goal of the Test Flight Day is to motivate organizations across the industry – Internet service providers, hardware makers, operating system vendors and web companies – to prepare their services for IPv6 to ensure a successful transition as IPv4 addresses run out. (source: World IPv6 day)
I'm happy to announce that aprs.fi is also taking part in this experiment, where IPv6 support is suddenly enabled for a large amount of high-profile web sites. Please note that IPv4 support will not be disabled during the experiment. I'm only adding the IPv6 AAAA DNS records for the main aprs.fi site, instead of having it as a separate entry on ipv6.aprs.fi.
If you find that aprs.fi stops working for you on Wednesday (give or take a day to allow for timezone differences), you'll also soon find a large number of other sites not working. Your network operator's network does not support web sites which provide service to both IPv4 and IPv6 users. According to statistics made by Wikimedia on the Wikipedia site, this will hit about 0.3% of users. If you happen to belong to this group, it's important for your network operator to realize that this will eventually become a permanent situation. Your network needs a little fixing.
You can prepare for the World IPv6 day by visiting one of these two testing sites: simple "OMG" version and verbose.
More information about the event on the World IPv6 day site. Happy IPv6 day, everyone!
Saturday, June 4, 2011
New map viewers record
Yesterday afternoon aprs.fi made a new record: there were well over 3000 users viewing the real-time map at the same time. Almost all of them were displaying the embedded map on the Copenhagen Suborbitals Launch Campaign June 2011 page. The guys built a rocket, capable of taking 1 man to the edge of the space, and did their first successful test flight yesterday. Awesome.
The launch campaing page's embedded APRS map shows the position of the Sputnik launch platform. The platform hosts a custom APRS tracker designed and built by OZ1EKD and OZ7HVO.
I was happy to see that aprs.fi did well with the larger amount of visitors. They were all looking at the same station, and the caching worked well - the database really didn't get any additional queries due to the amount of viewers. I was able to find some spots where additional optimization would be useful and could take down the CPU usage of the web service considerably. Here are some graphs from one of the two servers:
Sunday, May 15, 2011
aprs.fi caps & shirts
Just in time for the trip to Dayton Hamvention. :) I wonder if anyone would like to actually buy these?
Saturday, May 7, 2011
New logo graphics & Dayton Hamvention
I'm coming over to the US to visit Dayton Hamvention two weeks from now. See you there! If all goes well with the manufacturing, you should be able to recognize me from a nice aprs.fi branded hat & shirt.
For the clothing I had to create a new logo, which can already be seen on this blog and the Facebook page! I'm more of a musician and a programmer than a graphics artist, which is probably visible. :)
I've also been working with the new alerts and map filtering functions. They're quite large improvements and still need a lot of work on the user interface side, although the core functionality starts to take shape already. Stay tuned.
For the clothing I had to create a new logo, which can already be seen on this blog and the Facebook page! I'm more of a musician and a programmer than a graphics artist, which is probably visible. :)
I've also been working with the new alerts and map filtering functions. They're quite large improvements and still need a lot of work on the user interface side, although the core functionality starts to take shape already. Stay tuned.
Saturday, April 2, 2011
Map loading problems last night
A memory table containing all of the last position reports for the past 30 hours or so, used for quickly loading the real-time map view, filled up last night on one of the two front-end web servers at around 2010-04-01 23:35 UTC (2010-04-02 2:35 local time). I woke up, noticed the SMS alert and fixed the problem at 2010-04-02 6:50 UTC (9:50 local time).
While the table was full, replication was stopped, and about 50% of real-time map loads (without a station tracked, just an area view) resulted in no stations displayed. Also, 50% of page loads on the site last night returned a bit old data (the stopped frontend continued to display the 23:35 UTC state!).
No data was lost during the incident, APRS-IS data collection continued normally.
While the table was full, replication was stopped, and about 50% of real-time map loads (without a station tracked, just an area view) resulted in no stations displayed. Also, 50% of page loads on the site last night returned a bit old data (the stopped frontend continued to display the 23:35 UTC state!).
No data was lost during the incident, APRS-IS data collection continued normally.
Subscribe to:
Posts (Atom)




