test: rework unused_port_tcp_udp to try random candidate ports#5427
Draft
Watson1978 wants to merge 1 commit into
Draft
test: rework unused_port_tcp_udp to try random candidate ports#5427Watson1978 wants to merge 1 commit into
Watson1978 wants to merge 1 commit into
Conversation
0910773 to
1cc8953
Compare
unused_port_tcp_udp() collected 200 TCP ephemeral ports at once and returned the first one that could also be bound for UDP. On Windows, Hyper-V/WinNAT/HNS reserve "excluded port ranges" inside the ephemeral port range (49152-65535), the reservations are managed separately for TCP and UDP, and binding a port inside such a range fails with EACCES instead of EADDRINUSE. The TCP ephemeral allocator assigns ports sequentially and knows nothing about UDP exclusions, so the 200 candidates form a contiguous block that can land entirely inside a UDP-only excluded range. When that happens, every candidate fails the UDP bind check and test setup fails with: RuntimeError: can't find unused port test/helper.rb:94:in `unused_port_tcp_udp' test/helper.rb:81:in `unused_port' test/plugin/test_in_forward.rb:22:in `setup' This was captured on a windows-2025 GitHub Actions runner whose UDP exclusion ranges (netsh interface ipv4 show excludedportrange protocol=udp) were 54315-54414 and 54415-54514: exactly 200 consecutive ports, while the TCP exclusions were elsewhere (49691-49890). Instead, draw random candidate ports from a fixed high range and verify that each can be bound for both TCP and UDP, trying another candidate on failure. Because the candidates are independent draws, reserved or in-use ports only matter through their overall fraction of the range: contiguous reservation blocks cannot defeat the search, and the chance of exhausting the retry limit is negligible. The range is kept wide (10001 ports) so that reservation blocks, which come in units of about 100-200 ports, can only ever cover a small fraction of it. This also removes the need to hold 200 sockets at once, which the old implementation capped because of file descriptor limits, and the checks now bind both protocols explicitly on the same IPv4 wildcard address. Only the protocol: :all path changes; unused_port_tcp() and unused_port_udp() used by the :tcp/:tls/:udp paths are unchanged. Signed-off-by: Shizuo Fujita <fujita@clear-code.com>
1cc8953 to
ff8d8b5
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Which issue(s) this PR fixes:
Fixes #
What this PR does / why we need it:
unused_port_tcp_udp()collected 200 TCP ephemeral ports at once and returned the first one that could also be bound for UDP.On Windows, Hyper-V/WinNAT/HNS reserve "excluded port ranges" inside the ephemeral port range (49152-65535), the reservations are managed separately for TCP and UDP, and binding a port inside such a range fails with
EACCESinstead ofEADDRINUSE.The TCP ephemeral allocator assigns ports sequentially and knows nothing about UDP exclusions, so the 200 candidates form a contiguous block that can land entirely inside a UDP-only excluded range. When that happens, every candidate fails the UDP bind check and test setup fails with:
This was captured on a windows-2025 GitHub Actions runner whose UDP exclusion ranges (netsh interface ipv4 show excludedportrange protocol=udp) were 54315-54414 and 54415-54514: exactly 200 consecutive ports, while the TCP exclusions were elsewhere (49691-49890).
Instead, draw random candidate ports from a fixed high range and verify that each can be bound for both TCP and UDP, trying another candidate on failure. Because the candidates are independent draws, reserved or in-use ports only matter through their overall fraction of the range: contiguous reservation blocks cannot defeat the search, and the chance of exhausting the retry limit is negligible. The range is kept wide (10001 ports) so that reservation blocks, which come in units of about 100-200 ports, can only ever cover a small fraction of it.
This also removes the need to hold 200 sockets at once, which the old implementation capped because of file descriptor limits, and the checks now bind both protocols explicitly on the same IPv4 wildcard address.
Docs Changes:
N/A
Release Note:
N/A