in nifi, have http endpoint accepts post requests payloads varying 7kb 387kb or larger (up 4mb). goal have clustered implementation capable of handling approximately 10,000 requests per second. however, no matter if have nifi clustered 3 nodes or single instance, i've never been able average more 15-20 requests/second without jetty service returning 503 error. i've tried reducing time penalty , increasing number of maximum outstanding requests in standardhttpcontextmap. no matter try, whether on local machine or on remote vm, cannot impressive number of requests go through.
any idea why occurring , how fix this? when clustering, notice 1 node (not primary node) majority of work , think explains why throughput isn't higher clustered implementation.
no matter @ bulletin level, error in nifi-app.log:
2016-08-09 09:54:41,568 info [qtp1644282758-117] o.a.n.p.standard.handlehttprequest handlehttprequest[id=6e30cb0d-221f-4b36-b35f-735484e05bf0] sending service_unavailable response 127.0.0.1; request post 127.0.0.1 this same whether i'm running 2 processors (handlehttprequest , handlehttpresponse) processors or have general flow i'm routing on content, replacing text, , writing database or jms messaging system. can higher throughput (up 40 requests/sec) when i'm running web service without entire flow, still has ko rate of 90%, it's not better - still seems issue jetty service.
Comments
Post a Comment