Increment server failure on hostname lookup failures

[underwoo16]

This change increments server failure counter when MEMCACHED_HOST_LOOKUP_FAILURE occurs

This will allow the server to be ejected when persistent DNS issues occur to prevent constant retries

Sample test failure before change:

memcached/test/unit/memcached_test.rb:1224:in `test_dns_nxdomain_server'
     1221:
     1222:     # Hit second server and pass the limit
     1223:     key2 = "test_missing_server_3"
  => 1224:     begin
     1225:       cache.get(key2)
     1226:     rescue => e
     1227:       assert_equal Memcached::ServerIsMarkedDead, e.class
<Memcached::ServerIsMarkedDead> expected but was
<Memcached::HostnameLookupFailure>

[Copilot]

Pull Request Overview

This PR adds a test case to verify that server ejection is not currently working for DNS errors when connecting to memcached servers. The test demonstrates that servers experiencing hostname lookup failures should be ejected but currently are not being properly marked as dead.

• Adds a new test test_dns_nxdomain_server that exercises DNS error handling and server ejection
• Removes Unix socket server from test configuration to simplify the test setup
• Comments out UDP and domain socket memcached server initialization in setup

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
test/unit/memcached_test.rb	Adds DNS error test case and updates server configuration to exclude Unix socket
test/setup.rb	Comments out UDP and domain socket server initialization

[BrianGreenhill]

I suspect it will be hard to roll this change gradually in prod so glad to see a thorough test here.

Can we try it in review lab first?

[srt32]

Bad idea alert... ship the dep bump in a pr to review lab that also adds a non existent host to the mc config?