René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Using Redis sentinel, some commands 'send' hangs","articleBody":"**Describe the bug**\r\nWhen using Redis sentinel - 1 master, 2 duplicates, 3 sentinels - I can always connect to the (curent) writeable master.\r\n\r\nHowever, if I 'stop' that service, the next command hangs.\r\n\r\nThis may not be a bug, maybe it's proper use of Sentinel, but if this is the case the examples and/or docs should probably be updated. I can do a PR if that's the case.\r\n\r\n**To Reproduce**\r\nSteps to reproduce the behavior:\r\n1. Implement Redis Sentinel\r\n2. Connect to the current master\r\n3. Issue an 'lpop' command:\r\n```\r\n\tredisClient-\u003elpop(GetQueue().toStdString(), [\u0026webReply,\u0026bError](cpp_redis::reply\u0026 charRequest) {\r\n                      .... stuff ... \r\n\t\t});\r\n\tredisClient-\u003esync_commit(std::chrono::milliseconds(5000)); \r\n```\r\n\r\n4. It works fine.\r\n5. stop the current master. The slaves will fail over and promote a new master.\r\n6. The next 'lpop' command hangs down inside this routine:\r\n```\r\nclient::send(const std::vector\u003cstd::string\u003e \u0026redis_cmd, const reply_callback_t \u0026callback) {\r\n\tstd::lock_guard\u003cstd::mutex\u003e lock_callback(m_callbacks_mutex);\r\n\t__CPP_REDIS_LOG(info, \"cpp_redis::client attempts to store new command in the send buffer\");\r\n\tunprotected_send(redis_cmd, callback);\r\n\t__CPP_REDIS_LOG(info, \"cpp_redis::client stored new command in the send buffer\");\r\n\r\n```\r\n\r\n\r\n**Expected behavior**\r\nI would think issuing any serial commands to a connection that can't take it should result in a tacopie or cpp_redis error being thrown. \r\n\r\n**Possible workaround**\r\nMy routine that does the connect DOES immediately exits out:\r\n```\r\n\tLOG_INFO \u003c\u003c \"Connecting to HA Redis servers with \" \u003c\u003c iSentinels \u003c\u003c \"sentinels\";\r\n\tbool bDisconnect = false;\r\n\tredisClient-\u003econnect(\"ourcluster\", [\u0026bDisconnect](const std::string\u0026 host, std::size_t port, cpp_redis::connect_state status) {\r\n\t\tif (status == cpp_redis::connect_state::dropped) {\r\n\t\t\tLOG_INFO \u003c\u003c \"Client disconnected from \" \u003c\u003c host.c_str() \u003c\u003c \":\" \u003c\u003c port;\r\n```\r\n\r\nThe 'dropped' fires right off, so I could assign a bool value, and then in the outer loop, before every command check to see if we're disconnected - but that seems like a lot of potential boilerplate code.\r\n\r\nI guess I could also just try to reconnect inside the 'connect' callback? Is that the right way to do it? The code example didn't show this\r\n\r\n**Desktop (please complete the following information):**\r\n - OS: [Windows 10]\r\n - Browser [n/a]\r\n - Version [Current]\r\n\r\n","author":{"url":"https://github.com/jgwinner","@type":"Person","name":"jgwinner"},"datePublished":"2022-05-02T04:43:44.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":1},"url":"https://github.com/92/cpp_redis/issues/92"}