3xsd

Latest version: v0.0.26

Safety actively analyzes 681723 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 1 of 2

0.0.26

added:
3nsd GeoIP based name resolving(src ip based geo resolving)
3wdd scripts define on tunnel up/down/route-change to execute

optimized:
3wdd encryption logic: CFB/OFB/CTR padding is not required
3wdd encrypt codes, now with a Blowfish-ctr method, tunnel data transfer performance
is closed to raw non-encrypt tunnel. I believe that with a intel AES-NI supported CPU
(like XEON E3-1240/1270), AES-128-CTR can also do it

fixed:
3wdd In check_3wd() var: _fixed_metric bug fix

0.0.25

added:
3wdd blowfish-cbc/cfb/ctr encrypt method(it felt that's faster than aes-cbc/cfb/ecb?)

fixed:
3wdd bugs in route metric function: _route not in route_metric_fixed[], no key value error

0.0.24

added:
3wdd tunnel data relay option, see .conf, for systems not supporting tun interface creating
3wdd route metric, routing data through different path, depending on tunnel rtt(choose the best one)
3nsd global DNS Left-Right Range Resolve(LRRR), see white paper and .conf(experimental)

0.0.23

added:
3wdd config item 'io_mode' : block, default(block), non_block
block mode use 2 threads per tunnel with blocking io, bi-direction forwarding packages
non_block mode use 1 thread for tunnels per process with non-blocking io, epoll to handle udt&tun events
non_block mode take higher cpu consumtion than block mode, but will performance better in multi-tunnels env
on the other hand, block mode runs better in single-tunnel env

3zsd resolve dns ip of request hostname, then direct to it(transparent feature), through a special
config line: * = 100,0.0.0.0:0

fixed:
3wdd in wdd_idle_worker(), exclude worker_id==0 situation(master process should not recieve udt connection)
3wdd udt tunnel using epoll extremely slow issue, see https://github.com/zihuaye/3xsd/issues/2 for detail

optimized:
3wdd & 3zsd - using deque instead of list in zidles and udt_send_buf
3wsd _xHandler.send_out_all_headers() logical optimize

0.0.22

added:
3wdd tunnel on-the-fly compress function.
zlib & lzo support, lzo faster and lower CPU consumption, zlib get higher compress rate
should not use at already compressed data transfering
3wdd server side tunnel route support.

fixed:
3wdd struct.pack/unpack() byte order may be unmatched under different cpu arch issue, all set to "network"(big-endian).

0.0.21

optimized:
3wdd UDT & TUN handling loop major rewrite, cpu consumption rate decrease 0.6%.
Now UDT tunnel(no encrypt) performing 50%-60% speed of direct TCP connection with ZetaTCP,
and package lost rate remaining below 0.6%, while direct connection has 1.4%-3%.
However, UDT tunnel beats normal TCP connection without ZetaTCP, with 30%-50% outperforming.(v)
(Test CN-US WAN link with 150ms-280ms latency, through the always-jammed CUCN submarine cable)
And an encrypted UDT tunnel will has 50% performance decrease.

Page 1 of 2

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.