I have also tried with resurrect_ttl value of 0, still no chnage in the behaviour
2019/01/15 19:23:31 [warn] 65#0: *8992 [lua] mlcache.lua:780: get(): callback returned an error (table: 0x42aeb230) but stale value found in shm will be resurrected for 30s (resurrect_ttl), client: 100.124.0.10, server: kong, request: “GET /api/gateway/admin/healthcheck HTTP/1.1”, host: “health.xxxx-xxx-kongsrvr.us-xxxx-2.csp.xxxx.com”
2019/01/15 19:23:31 [info] 65#0: *8992 [lua] csp_session_client.lua:236: getSessionWithClientCrt(): Cached Session data type string, client: 100.124.0.10, server: kong, request: “GET /api/gateway/
The Kong singletons cache is kongs main cache, and if you look in the core code you will see where they do initialize it with a resurrect_ttl of 30 seconds. You will need to declare your own mlcache if you don’t want that functionality.
When I add lua_shared_dict to the nginx-kong.conf an drestart Kong for some reason Kong overrides my change and set the nginx-kong.conf back to the old unchanged version. Have you ever encountered this issue?