Error for automatic filter list updates

[Caro]

For me EasyList does sometimes noch get automatically updated. I get an error message right next to the filter list?

Failed to load resource: net::ERR_NAME_NOT_RESOLVED
https://easylist-downloads.adblockplus.org/easylistgermany+easylist.txt?add…&platformVersion=47.0.2526.111&lastVersion=201601180921&downloadCount=4%2B 

Adblock Plus: Downloading URL https://easylist-downloads.adblockplus.org/easylistgermany+easylist.txt failed (synchronize_connection_error)
Download address: https://easylist-downloads.adblockplus.org/easylistgermany+easylist.txt?add…&platformVersion=47.0.2526.111&lastVersion=201601180921&downloadCount=4%2B
Channel status: -1
Server response: 0

[matze]

Looks like there are some issues with the DNS "balancing", again. I suppose it's related to the currently ongoing migration to a new balancing scheme and hope it will vanish afterwards. We'll see.

[matze]

The issue's severity seems to increase, there are more frequent but still quite short events. We now get reports from users of other browsers, i.e. Chrome, as well:

This webpage is not available

DNS_PROBE_FINISHED_NXDOMAIN

The server at easylist-downloads.adblockplus.org can't be found, because the DNS lookup failed.

I did not yet manage to capture any malformed records. The logs of the interval-based monitoring I've set up before did not catch any bad records yet. It seems like the respective browser's cache prolongs the use of empty ones occasionally delivered by our DNS provider, but unfortunately also the manual checks after noticing (on affected machines) do not reveal any actual issues either.

[matze]

Clients with 5 cosectuive errors Apr/04 - May/02

[matze]

The newly established traffic management (#286) seems to improve the situation. Together with more resources to handle traffic peaks introduced recently issues like this one should become quite rare. Still, it will most likely re-occur whenever one of our up-links is over-loaded.

Our client applications are build to handle these outages and usually simply re-try the operation later, hence this is not considered being a severe issue. Nonetheless we are working on improved monitoring (i.e. #3981), in order to earlier discover and tackle bottlenecks in the future.

---