Codecov Report

Merging #160 into dev will decrease coverage by -0.13%.
The diff coverage is 0%.

@@             Coverage Diff              @@
##                dev     #160      +/-   ##
============================================
- Coverage     29.77%   29.65%   -0.13%     
+ Complexity     1251     1247       -4     
============================================
  Files            97       97              
  Lines         23303    23305       +2     
  Branches       3871     3871              
============================================
- Hits           6939     6910      -29     
- Misses        15014    15043      +29     
- Partials       1350     1352       +2

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6e8dc59...c086691. Read the comment docs.

I tried out our test case and it fixes the cancellation issue when multiSubnetFailover is enabled. It no longer hangs and the cancellation happens correctly. Very nice.

Skimming through the implementation, it looks like the approach is to close the socket channel and reconnect it as a regular socket. That should be be fine though it does lead to the situation that all multi subnet enabled connections will create two socket connections to the server (the first being closed). It'd be marginally slower than directly using the channel but it shouldn't be a big deal as it's just the establishment of the TCP socket. Also, I don't think it's an issue on the server to have the first connection closed as the driver is already doing that to the rest of the resolved addresses anyway.

So looks good to me.

FYI, the commit message has a typo: Convert socket adaptor to socket #160. Adaptor should be spelled adapter). Looks like it's already been merged though ...

@sehrope Thank you for getting back to us. We are glad to know you are satisfied with this fix.
Forgive me about the typo. English isn't my first language, Java is :)