Conversation
When the MultiSubnetFailover=Yes property is added to the connection string, the TCP connection should be attempted for each resolved IP address in parallel rather than in sequence. This creates a race where the first connection to be established wins and becomes the target server. https://learn.microsoft.com/en-us/sql/relational-databases/native-client/features/sql-server-native-client-support-for-high-availability-disaster-recovery?view=sql-server-ver15#connecting-with-multisubnetfailover
srkaj
force-pushed
the
multi-subnet-failover
branch
from
April 11, 2025 13:05
36c051a to
2732e81
Compare
Now better implemented in sql_brower with tokio, async_std and smol.
srkaj
force-pushed
the
multi-subnet-failover
branch
from
April 11, 2025 13:16
2732e81 to
0d1660f
Compare
joelparkerhenderson
added a commit
to mssql-rust/mssql-rust
that referenced
this pull request
Aug 29, 2026
…ects Mirrors tiberius-rs/tiberius#378. Adds `Config::multi_subnet_failover`/`Config::get_multi_subnet_failover` (also parsed from the `MultiSubnetFailover`/`multiSubnetFailover` ADO.NET and JDBC connection-string property, mirroring the existing `readonly`/ `ApplicationIntent` parsing pattern). When enabled, `SqlBrowser:: connect_named` (all three backends: tokio, async-std, smol) races a connection attempt against every one of the resolved addresses concurrently via `FuturesUnordered`, returning as soon as any succeeds, instead of trying them one at a time. This targets SQL Server Always On availability group listeners, whose DNS name can resolve to addresses on multiple subnets where only one is reachable at a time - trying them sequentially can take as long as the TCP connect timeout per dead address before reaching the live one. The per-address connect logic (SSRP/MS-SQLR instance-name resolution + TCP connect) was factored out into a `connect_addr` helper in each backend so both the sequential (default) and concurrent (multi_subnet_failover) branches share one implementation - this mirrors the existing (non-multi_subnet_failover) control flow exactly for the default case, just moved into a helper function, so its behavior for existing callers is unchanged. `Config::multi_subnet_failover` only affects the `SqlBrowser:: connect_named` code path (used for SQL Server named-instance discovery via the sql-browser-* features); it has no effect on `Client::connect`, which takes an already-established `TcpStream` - documented on both. Added unit tests for the config plumbing (default, explicit set, and ADO.NET/JDBC connection-string parsing, including an invalid-value error case). Live testing of the actual connect_named race behavior would need a real SQL Server Always On listener spanning multiple subnets (or at minimum a Windows SQL Browser service for the existing named-instance tests), neither available in this environment - same gap upstream's PR has (it shipped with no tests at all). The refactor was verified not to change behavior for the default (non-failover) path by inspection: the sequential branch is a direct, mechanical extraction of the pre-existing loop into `connect_addr`, unchanged statement-for-statement. Verified: - cargo check: default, --features=rustls, --features=vendored-openssl, --no-default-features, and each of sql-browser-tokio/async-std/smol - cargo clippy --all-targets -- -D warnings, for default and each sql-browser-* feature - cargo fmt --check - cargo test --lib (169 passed) - Live integration tests over rustls (unaffected code paths): 294 passed (92 bulk + 202 query) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0156Di1tRRLsJK8ctU1AmAJr
joelparkerhenderson
added a commit
to mssql-rust/mssql-rust
that referenced
this pull request
Aug 30, 2026
…ects Mirrors tiberius-rs/tiberius#378. Adds `Config::multi_subnet_failover`/`Config::get_multi_subnet_failover` (also parsed from the `MultiSubnetFailover`/`multiSubnetFailover` ADO.NET and JDBC connection-string property, mirroring the existing `readonly`/ `ApplicationIntent` parsing pattern). When enabled, `SqlBrowser:: connect_named` (all three backends: tokio, async-std, smol) races a connection attempt against every one of the resolved addresses concurrently via `FuturesUnordered`, returning as soon as any succeeds, instead of trying them one at a time. This targets SQL Server Always On availability group listeners, whose DNS name can resolve to addresses on multiple subnets where only one is reachable at a time - trying them sequentially can take as long as the TCP connect timeout per dead address before reaching the live one. The per-address connect logic (SSRP/MS-SQLR instance-name resolution + TCP connect) was factored out into a `connect_addr` helper in each backend so both the sequential (default) and concurrent (multi_subnet_failover) branches share one implementation - this mirrors the existing (non-multi_subnet_failover) control flow exactly for the default case, just moved into a helper function, so its behavior for existing callers is unchanged. `Config::multi_subnet_failover` only affects the `SqlBrowser:: connect_named` code path (used for SQL Server named-instance discovery via the sql-browser-* features); it has no effect on `Client::connect`, which takes an already-established `TcpStream` - documented on both. Added unit tests for the config plumbing (default, explicit set, and ADO.NET/JDBC connection-string parsing, including an invalid-value error case). Live testing of the actual connect_named race behavior would need a real SQL Server Always On listener spanning multiple subnets (or at minimum a Windows SQL Browser service for the existing named-instance tests), neither available in this environment - same gap upstream's PR has (it shipped with no tests at all). The refactor was verified not to change behavior for the default (non-failover) path by inspection: the sequential branch is a direct, mechanical extraction of the pre-existing loop into `connect_addr`, unchanged statement-for-statement. Verified: - cargo check: default, --features=rustls, --features=vendored-openssl, --no-default-features, and each of sql-browser-tokio/async-std/smol - cargo clippy --all-targets -- -D warnings, for default and each sql-browser-* feature - cargo fmt --check - cargo test --lib (169 passed) - Live integration tests over rustls (unaffected code paths): 294 passed (92 bulk + 202 query) Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0156Di1tRRLsJK8ctU1AmAJr
This was referenced Sep 1, 2026
MattJackson
added a commit
that referenced
this pull request
Sep 3, 2026
Integrate upstream PR #378 (resolves #337). Add a `multi_subnet_failover` option to `Config` with a setter/getter and ADO/JDBC connection-string parsing of the `MultiSubnetFailover` keyword (via the shared `ConfigString` trait, so it works for both string formats). When enabled, the SQL Browser `connect_named` paths (tokio, async-std, smol) attempt connections to every resolved address in parallel using `FuturesUnordered` and return the first stream to succeed; otherwise they fall back to the existing sequential behaviour. The per-address connect logic is factored into a `connect_addr` helper, and the first observed error is now preserved instead of being masked by a generic "not found". Adds unit tests for keyword parsing and end-to-end `from_ado_string` propagation.
MattJackson
added a commit
that referenced
this pull request
Sep 4, 2026
Integrate upstream PR #378 (resolves #337). Add a `multi_subnet_failover` option to `Config` with a setter/getter and ADO/JDBC connection-string parsing of the `MultiSubnetFailover` keyword (via the shared `ConfigString` trait, so it works for both string formats). When enabled, the SQL Browser `connect_named` paths (tokio, async-std, smol) attempt connections to every resolved address in parallel using `FuturesUnordered` and return the first stream to succeed; otherwise they fall back to the existing sequential behaviour. The per-address connect logic is factored into a `connect_addr` helper, and the first observed error is now preserved instead of being masked by a generic "not found". Adds unit tests for keyword parsing and end-to-end `from_ado_string` propagation.
MattJackson
added a commit
that referenced
this pull request
Sep 6, 2026
Integrate upstream PR #378 (resolves #337). Add a `multi_subnet_failover` option to `Config` with a setter/getter and ADO/JDBC connection-string parsing of the `MultiSubnetFailover` keyword (via the shared `ConfigString` trait, so it works for both string formats). When enabled, the SQL Browser `connect_named` paths (tokio, async-std, smol) attempt connections to every resolved address in parallel using `FuturesUnordered` and return the first stream to succeed; otherwise they fall back to the existing sequential behaviour. The per-address connect logic is factored into a `connect_addr` helper, and the first observed error is now preserved instead of being masked by a generic "not found". Adds unit tests for keyword parsing and end-to-end `from_ado_string` propagation.
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
2 participants
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.
This builds on PR #357 to fix #337.
MultiSubnetFailoveris a recongnized configuration flag.Config.