Site connector going down daily after upgrade

Products

• Symantec ZTNA

Issue/Introduction

ZTNA setup and working fine with 2.9.4 connector deployed via docker compose.

After upgrading to the 2.10.13 connector deployed via docker compose, the connector disconnects almost daily.

ZTNA connector logs showing exception within Websocket connections as shown below:

time="2024-01-10T09:06:37Z" level=fatal msg="runtime error: invalid memory address or nil pointer dereference" Ticket=2bd01699-e0aa-4538-a9b6-09864211f9cc connectorVersion=2.10.13+5072 error_trace="goroutine 562 [running]:\nruntime/debug.Stack()\n\t/usr/local/go/src/runtime/debug/stack.go:24 
+0x65\nbitbucket.org/accezz-io/connector/logging.CapturePanicAndWait.func1()\n\t/home/circleci/project/logging/loggers.go:39 +0x51\npanic({0xb32d80, 0x1142380})\n\t/usr/local/go/src/runtime/panic.go:884 +0x212\nbitbucket.org/accezz-io/connector/connection_handlers.(*WebsocketTracer).gotConnFunc.func1({{0x0?, 0x0?}, 0x40?, 0x17?, 0x4?})\n\t/home/circleci/project/connection_handlers/connection_handlers_helpers.go:147 +0x93\nbitbucket.org/accezz-io/websocket.(*Dialer).DialContextWithRequest(0xc000b6a6e0?, {0xd25fa8, 0xc000c0bd70}, {0xc000771700, 0x3d}, 0xc0004eb400)\n\t/home/circleci/project/vendor/bitbucket.org/accezz-io/websocket/client.go:498 +0x8ea\nbitbucket.org/accezz-io/connector/connection_handlers.(*WebsocketConnectionHandler).ConnectionCreatorInterceptor(0xc000b6a640, {0xc000771700, 0x3d}, 0xc0004eb200, 0x11f1001?)\n\t/home/circleci/project/connection_handlers/websocket_connection_handler.go:104 +0x3e6\nbitbucket.org/accezz-io/connector/connection_handlers.(*BaseHttpConnectionHandler).handleStart(0xc0001bab40, 0xc000b6a600)\n\t/home/circleci/project/connection_handlers/http_connection_handler_base.go:206 +0xcba\nbitbucket.org/accezz-io/connector/connection_handlers.(*BaseHttpConnectionHandler).Start(0xc0001bab40, 0xc000136d28?)\n\t/home/circleci/project/connection_handlers/http_connection_handler_base.go:112 +0x1e\nbitbucket.org/accezz-io/connector/connection_handlers.(*WebsocketConnectionHandler).Start(0xc000b6a640, 0xc000049000?)\n\t/home/circleci/project/connection_handlers/websocket_connection_handler.go:132 +0x16a\nbitbucket.org/accezz-io/connector/channel_consumers.(*ChannelConsumer).RequestHandler.func1()\n\t/home/circleci/project/channel_consumers/channel_consumer.go:121 +0x26\nbitbucket.org/accezz-io/connector/logging.CapturePanicAndWait(0x0?)\n\t/home/circleci/project/logging/loggers.go:48 +0x3e\ncreated by bitbucket.org/accezz-io/connector/channel_consumers.(*ChannelConsumer).RequestHandler\n\t/home/circleci/project/channel_consumers/channel_consumer.go:120 +0x5aa\n" PID=17
{"id":"####-####-####-####-####","date":"2024-01-10T09:06:37.128416189Z","tenantId":"Not available","level":"WARN","message":"Unable to signal /home/luminate//connector-stable with child exited(17), ignoring","requestId":"Not available","serviceName":"Not available","funcName":"bitbucket.org/accezz-io/connector/entrypoint_rpc.(*EntrypointRPCServer).broadcastSignal","additionalVars":{"connectorVersion":"2.10.13+5072"},"error":"os: process already finished"}

Manually restarting the connector or docker instance allows the solution to work again, for up to a day.

Environment

ZTNA / SAC.

Connector setup via docker compose.

Cause

Bug with websocket connections in 2.9.4.

Resolution

Upgrade to version 2.10.14 of the connector or greater.