From 55747935a39632ada8701cad61f12b69c065b7c8 Mon Sep 17 00:00:00 2001
From: Nedyalko Dyakov <1547186+ndyakov@users.noreply.github.com>
Date: Tue, 19 Aug 2025 15:39:20 +0300
Subject: [PATCH] Update redis.go

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
---
 redis.go | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/redis.go b/redis.go
index e5cca4af..27bd6b7e 100644
--- a/redis.go
+++ b/redis.go
@@ -930,7 +930,16 @@ func NewClient(opt *Options) *Client {
 		if err != nil {
 			internal.Logger.Printf(context.Background(), "hitless: failed to initialize hitless upgrades: %v", err)
 			if opt.HitlessUpgradeConfig.Mode == hitless.MaintNotificationsEnabled {
-				// panic so we fail fast without breaking existing clients api
+				/*
+					Design decision: panic here to fail fast if hitless upgrades cannot be enabled when explicitly requested.
+					We choose to panic instead of returning an error to avoid breaking the existing client API, which does not expect
+					an error from NewClient. This ensures that misconfiguration or critical initialization failures are surfaced
+					immediately, rather than allowing the client to continue in a partially initialized or inconsistent state.
+					Clients relying on hitless upgrades should be aware that initialization errors will cause a panic, and should
+					handle this accordingly (e.g., via recover or by validating configuration before calling NewClient).
+					This approach is only used when HitlessUpgradeConfig.Mode is MaintNotificationsEnabled, indicating that hitless
+					upgrades are required for correct operation. In other modes, initialization failures are logged but do not panic.
+				*/
 				panic(fmt.Errorf("failed to enable hitless upgrades: %w", err))
 			}
 		}