OK, it’s been 3 days since I updated to v2.13.1. I build from source as per docs.
The post mentioned fixing or changes to DTLS fingerprinting.
Is there anything we operators should notice like more traffic, more connections, less timeouts because I see none of that. Since RU was blocking by DTLS fingerprint I expect that there would be more connections.
I did not use -covertdtls-config randomizemimic since the docs say it is the default even though it is not clear as 1 user pointed out.
@BobbyB I just want to hijack your post re v2.13.1 to say the update for OpenWrt is in the works. If any OpenWrt Snowflake proxy users out there would like to support its deployment sooner rather than later the PR is here on GitHub.
\end hijack
No problem.
Out of curiosity how much traffic and/or connections can these devices carry. Guess it depends how powerful the device is but in general.
My memory of OpenWrt was using it with a home router. This was a long time ago.
Out of curiosity how much traffic and/or connections can these devices carry
Well I run Snowflake standalone proxy on my home router on the full ephemeral ports range and no client capacity limit. 7 concurrent connections (incl. IPv6) seems to be the maximum number I get allocated* and memory limit is around 230MB when maxed out - i.e. when throughput is very high.
I was upgrading my old router and went for OpenWrt’s own product; the OpenWrt One which has 1GB of RAM. Many home routers which can run OpenWrt have less than this, but even so should be able to run Snowflake proxy, although it may be necessary to specify a capacity limit.
*From reading comments here from the Snowflake team, it seems the number of Snowflake proxies out there available to clients is the important factor for censorship circumvention, not the capacity of any one proxy. Also, I understand client allocation by a broker depends on the physical location of a Snowflake proxy in relation to requesting clients - i.e. if you are near Russia/Iran etc you may see higher connection numbers.