Release v0.6.0 - Structured Data, Better Config, Agent Authn, and more! · natlas/natlas

Server changes

Administrators can now require that agents be authenticated in order to get and submit work.
Administrators can assign tags to scope ranges, which will then get automatically applied to the resultant scans
Administrators can now delete individual scan results as well as entire host results.
Setup script now automatically generates a random secret key if one is not already set.
Most administration pages have been improved by loading the tables into DataTables, which allow for sorting, searching, and pagination of results. This appears to improve load time, as well.
Users can now register their agents for scanning
Any authenticated user can request that an address be rescanned, which will get inserted into a queue that gets handed out prior to PRNG targets being assigned again.
Nmap XML data gets parsed and pulled into semi-structured elasticsearch documents now, paving the way for more advanced queries and statistics about your networks.
jQuery and Bootstrap versions got a major upgrade
New updated view to move away from the ugly raw nmap data on the screen
Host History page has been updated to provide a more streamlined set of information as well as display entries for when the host was detected as down, timed out, or otherwise didn't have results for it.
Scan results can now be exported as json documents, exactly the way the data is stored in elasticsearch.
Many many html errors were corrected, which should result in a more stable viewing experience.

Authentication ID and Token support
Option to ignore certificate errors when connecting to the natlas server via NATLAS_IGNORE_SSL_WARN environment variable. (This is useful for working with self-signed certificates, alternatively you can (and should probably) install the issuing CA as a trusted issuer on your agents).
Nmap scans that timeout are now being properly cleaned up after
Agents now submit status updates (and raw output files) for every host they get work for, whether that host is up, down, timed out, or is up but has no ports open.
Agents can be configured to retry any number of times before giving up on submitting data - this can be useful in edge cases where your agents have collected some data that the server doesn't understand, as it prevents the agent/server from getting into an infinite loop of trying to submit data that doesn't work.
Agents no longer request extra work before they are ready for it

In certain currently-unknown circumstances, when aquatone closes it doesn't clean up the chromium-browser child processes, which can eventually lead to system stability problems via resource exhaustion. This should hopefully be resolved upstream soon.
Structured port information can't be directly searched via the search box yet, the search box is using query_strings which do not support nested objects.
I am still not using a proper api model, but I promise that I'm actively thinking about it and figuring out how to move towards it entirely.

In the time it took me to develop and debug the necessary features for this release, I've learned that basically everything I've done so far is wrong and could have been done way better.
In that same time, I've watched Hackers a dozen times, and listened to Halcyon On and On for at least 8 hours.