policy-list with ~30K rules times out (haproxy resets connection after 50 secs timeout)

[Senthilnathan Murugappan]

neutron-client(Haproxy)->neutron-server
18:45:07.978128 IP 10.84.7.41.39732 > 10.84.7.41.9697: Flags [S], seq 4063197695, win 43690, options [mss 65495,sackOK,TS val 46569120 ecr 0,nop,wscale 7], length 0

Neutron-server->cfgm(haproxy)
18:45:08.135807 IP 10.84.7.41.59998 > 10.84.7.41.8082: Flags [S], seq 2337626520, win 43690, options [mss 65495,sackOK,TS val 46569160 ecr 0,nop,wscale 7], length 0
18:45:27.635799 IP 10.84.7.41.8082 > 10.84.7.41.59998: Flags [F.], seq 18024643, ack 589, win 351, options [nop,nop,TS val 46574035 ecr 46574035], length 0

Though we get the reply from api-server in about 20 secs. The below pformat(policy_dicts) alone takes about 36 secs.

        LOG.debug(
            "get_policys(): filters: " + pformat(filters) + " data: "
            + pformat(policy_dicts))

Reset sent from haproxy to neutron-server on 50 sec timeout
18:45:57.980912 IP 10.84.7.41.39732 > 10.84.7.41.9697: Flags [R.], seq 4063, ack 1, win 342, options [nop,nop,TS val 46581621 ecr 46569120], length 0

[Senthilnathan Murugappan (msenthil)]

Would be better if we could support sending chunks of data.