Repository navigation
Replies: 2 comments
|
For reliable attribution, this needs information from the operating system, not just another field in the DNS log. A standard DNS request does not carry the originating application's PID or executable name; its transaction ID identifies the DNS exchange, not the process (message format). The current query logger records the client IP and query/result information, but has no original-process field. There are OS-level tools for the question you're trying to answer:
Trying to map the proxy's incoming socket back to a PID can be racy for short-lived UDP sockets, and if an application delegates to a shared system resolver, it may only identify that resolver. Separate proxy instances/ports can label traffic you explicitly route, but do not recover the missing application identity for the rest. Which OS are you using? That would help narrow down a useful tracer and its coverage. |
|
@Pavel-Shcherbo Pardon me for not adding crucial info. I completely forgot this package is available also for non *nix OSes. |
Uh oh!
There was an error while loading. Please reload this page.
howdy
I've been wondering whether it could be possible to implement monitoring processes names/PIDs by dnscrypt-proxy?
I use it primary as a localhost dns cache, and sometimes I wonder what the heck made such a weird request I see in monitor dashboard.
The only solution I can think of is launching multiple instances of dnscryptproxy, and point certain software through different ports, but it still leaves too much space for guessing when it goes to all system wide packages working in background, that would use default instance.
I don't even have a slight idea what type of software I could use for this purpose, but I suspect all the hassle may not be worth the effort.
Maybe somebody already thought about it?
All reactions