Hi all, We have been testing multihoming using lksctp, expected outcome was immediate primary path failure identification and change over to secondary address at least within 2 seconds. With the following sctp parameters, it took 63 seconds to recover. net.sctp.rto_initial = 3000 net.sctp.rto_min = 1000 net.sctp.rto_max = 60000 net.sctp.path_max_retrans = 5 net.sctp.hb_interval = 30000 We can reason-out this observation according to total primary path RTO expiration time calculation. Since it was poor performance we have changed those parameters as net.sctp.rto_min = 40 net.sctp.rto_max = 800 net.sctp.path_max_retrans = 4 it took 1 second to change over to secondary, but time to time fluctuation is observed and at the time of fluctuation, packets were tried to send through unavailable primary path and then re-transmitted to secondary. Are we missing any parameters related to multihoming ? lksctp used: lksctp-tools-1.0.10-3.el5.rfx.x86_64.rpm kernal: Linux x86_64 2.6.39-4.el5.elrepo Below is the whole sctp parameters configured. net.sctp.rto_initial = 3000 net.sctp.rto_min = 1000 net.sctp.rto_max = 60000 net.sctp.valid_cookie_life = 60000 net.sctp.max_burst = 15 net.sctp.association_max_retrans = 10 net.sctp.sndbuf_policy = 1 net.sctp.rcvbuf_policy = 1 net.sctp.path_max_retrans = 5 net.sctp.max_init_retransmits = 8 net.sctp.hb_interval = 30000 net.sctp.cookie_preserve_enable = 1 net.sctp.rto_alpha_exp_divisor = 3 net.sctp.rto_beta_exp_divisor = 2 net.sctp.addip_enable = 0 net.sctp.prsctp_enable = 1 net.sctp.sack_timeout = 1 net.sctp.sctp_mem = 11377184 11502912 11754368 net.sctp.sctp_rmem = 114096 11361500 114194304 net.sctp.sctp_wmem = 114096 1116384 114194304 net.sctp.auth_enable = 0 net.sctp.addip_noauth_enable = 0 net.sctp.addr_scope_policy = 1 net.sctp.rwnd_update_shift = 4 Thanks and regards, P.Nirosan. -- To unsubscribe from this list: send the line "unsubscribe linux-sctp" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
