From bf66adc86b104ff82793a89a4aa64e81e0a03ecc Mon Sep 17 00:00:00 2001
From: Yuri Smirnov <tycoooon@gmail.com>
Date: Wed, 29 Apr 2026 12:48:44 +0300
Subject: [PATCH 1/2] Retry once on stale-connection ConnectionError in
 Pool#with
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Pool checkouts can hand out connections whose server-side end has
been closed while idle (CH idle_connection_timeout / tcp_keep_alive_timeout,
LB / firewall idle drops, server restart). The first read on such a
socket sees recv() == 0, which clickhouse-cpp surfaces as

  std::system_error(errno=0, "closed")

and we wrap as ClickhouseNative::ConnectionError("closed: Success").
The pool already discards the dead client (`discard_current_connection`
in the rescue) so the next checkout is fresh — but we used to re-raise
the error to the caller, who'd see a noisy "closed: Success" failure
on the very first operation after their pooled client went stale.

Now Pool#with discards-and-retries once on ConnectionError. If the
retry also fails (e.g. server actually down), we raise. Other error
classes (ServerError, ProtocolError, EncoderError, etc.) are still
not retried — those are real and re-running could mask logic bugs
or, for ServerError on a half-completed write, double-execute.

Three new specs cover the path: success on retry, raise after the
second failure, and no retry on non-ConnectionError.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 lib/clickhouse_native/pool.rb  | 23 ++++++++++++++++++----
 spec/clickhouse_native_spec.rb | 35 ++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+), 4 deletions(-)

diff --git a/lib/clickhouse_native/pool.rb b/lib/clickhouse_native/pool.rb
index 602eb55..9ab2a51 100644
--- a/lib/clickhouse_native/pool.rb
+++ b/lib/clickhouse_native/pool.rb
@@ -29,11 +29,26 @@ def initialize(host:, port:, database: "default", user: "default", password: "",
     # session settings), producing misleading log lines and re-raises in
     # unrelated code. A fresh socket + handshake is cheap relative to
     # debugging that.
+    #
+    # ConnectionError gets one automatic retry: pooled connections that
+    # have been idle long enough for the server / an LB to FIN them
+    # surface as "closed" on the very next recv (errno 0, message
+    # "closed: Success"). Discarding and re-checking out lands a fresh
+    # socket and the operation succeeds. The retry only triggers when
+    # the dead-connection error fired before any data was sent, so
+    # write operations don't risk double-execution from this path.
     def with
-      @pool.with do |client|
-        yield client
-      rescue
-        @pool.discard_current_connection(&:close)
+      attempts = 0
+      begin
+        @pool.with do |client|
+          yield client
+        rescue
+          @pool.discard_current_connection(&:close)
+          raise
+        end
+      rescue ConnectionError
+        attempts += 1
+        retry if attempts == 1
         raise
       end
     end
diff --git a/spec/clickhouse_native_spec.rb b/spec/clickhouse_native_spec.rb
index 6e31cdb..586522f 100644
--- a/spec/clickhouse_native_spec.rb
+++ b/spec/clickhouse_native_spec.rb
@@ -477,6 +477,41 @@
     expect(pool.database).to eq("default")
   end
 
+  describe "ConnectionError retry" do
+    it "retries once after a stale-connection ConnectionError" do
+      attempts = 0
+      result = pool.with do |client|
+        attempts += 1
+        raise ClickhouseNative::ConnectionError, "closed: Success" if attempts == 1
+        client.query_value("SELECT 1")
+      end
+      expect(result).to eq(1)
+      expect(attempts).to eq(2)
+    end
+
+    it "raises after the second ConnectionError" do
+      attempts = 0
+      expect do
+        pool.with do
+          attempts += 1
+          raise ClickhouseNative::ConnectionError, "closed: Success"
+        end
+      end.to raise_error(ClickhouseNative::ConnectionError)
+      expect(attempts).to eq(2)
+    end
+
+    it "does not retry non-connection errors" do
+      attempts = 0
+      expect do
+        pool.with do
+          attempts += 1
+          raise ClickhouseNative::ServerError, "boom"
+        end
+      end.to raise_error(ClickhouseNative::ServerError)
+      expect(attempts).to eq(1)
+    end
+  end
+
   describe "settings:" do
     it "applies Integer settings to every client on checkout" do
       p = described_class.new(**CH_KWARGS, pool_size: 2, settings: { max_threads: 7 })

From b82f91b27639570090cce561bd3324701dfd24ec Mon Sep 17 00:00:00 2001
From: Yuri Smirnov <tycoooon@gmail.com>
Date: Wed, 29 Apr 2026 12:52:45 +0300
Subject: [PATCH 2/2] Bump version 0.7.0 -> 0.8.0

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---
 lib/clickhouse_native/version.rb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/clickhouse_native/version.rb b/lib/clickhouse_native/version.rb
index 7ae34ac..e5d17a0 100644
--- a/lib/clickhouse_native/version.rb
+++ b/lib/clickhouse_native/version.rb
@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module ClickhouseNative
-  VERSION = "0.7.0"
+  VERSION = "0.8.0"
 end