Hello! I am writing a small load testing tool in python for a payment solutions company, where I'm basically parsing logs containing incoming xmlrpc calls from various e-stores. There are maybe 10.000+ e-stores, and for arguments sake we can say that maybe 100 of them may be active concurrently and sending requests to the server.
Each e-store is identified by an ID. So what I am doing is parsing the logs, extracting the calls and timestamps, and replaying the same traffic patterns in 1.0x speed, or any other speed factor. I would also like to simulate the connection dynamics so that every active e-store has its own connection. Since there can be 10.000 e-stores, I would like to reuse say 100 worker threads in a pool, where each thread would be assigned an e-store ID, and if it has been idle for a while it could get assigned a new one. The router process should be able to set e-store ID for a thread that has been idle for a certain amount of time and route it that way. I'm fumbling a bit regarding which socket type to use, and how to set the ID on the fly, and I'm wondering if anyone has any ideas and if what I wrote makes any sense so far :) Regards, /Adde
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
