TCP ACK loop with 2.6.33-rc6

TCP ACK loop with 2.6.33-rc6

[Simon Arlott]

[-- Attachment #1: Type: text/plain, Size: 3533 bytes --]

Packet capture attached. Is it possible to tell which side is at fault?
They are both responding to each other's ACK with another ACK at 400pps.

10:32:28.991111 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331059 251057709>
10:32:28.995046 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:28.995092 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331063 251057709>
10:32:29.009052 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:29.009112 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331077 251057709>
10:32:29.010046 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:29.010093 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331078 251057709>
10:32:29.014039 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:29.014087 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331082 251057709>
10:32:29.018052 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:29.018129 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331086 251057709>
10:32:29.021053 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559
10:32:29.021097 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331089 251057709>

The latency to 2001:500:b::1 is about 45ms (another ACK arrives before
the response to the previous ACK could have been received).

The trailing 0 bytes on packets going to 2001:8b0:ffea:0:205:b4ff:fe12:530
have been added by my ISP to avoid a Cisco bug.

net.ipv4.tcp_timestamps = 1
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_sack = 1
net.ipv4.tcp_retrans_collapse = 1
net.ipv4.tcp_syn_retries = 5
net.ipv4.tcp_synack_retries = 5
net.ipv4.tcp_max_orphans = 131072
net.ipv4.tcp_max_tw_buckets = 180000
net.ipv4.tcp_keepalive_time = 7200
net.ipv4.tcp_keepalive_probes = 9
net.ipv4.tcp_keepalive_intvl = 75
net.ipv4.tcp_retries1 = 3
net.ipv4.tcp_retries2 = 15
net.ipv4.tcp_fin_timeout = 60
net.ipv4.tcp_syncookies = 1
net.ipv4.tcp_tw_recycle = 0
net.ipv4.tcp_abort_on_overflow = 0
net.ipv4.tcp_stdurg = 0
net.ipv4.tcp_rfc1337 = 0
net.ipv4.tcp_max_syn_backlog = 1024
net.ipv4.tcp_orphan_retries = 0
net.ipv4.tcp_fack = 1
net.ipv4.tcp_reordering = 50
net.ipv4.tcp_ecn = 1
net.ipv4.tcp_dsack = 1
net.ipv4.tcp_mem = 82752	110336	165504
net.ipv4.tcp_wmem = 262144	1048576	4194304
net.ipv4.tcp_rmem = 262144	1048576	4194304
net.ipv4.tcp_app_win = 31
net.ipv4.tcp_adv_win_scale = 2
net.ipv4.tcp_tw_reuse = 0
net.ipv4.tcp_frto = 2
net.ipv4.tcp_frto_response = 0
net.ipv4.tcp_low_latency = 0
net.ipv4.tcp_no_metrics_save = 0
net.ipv4.tcp_moderate_rcvbuf = 1
net.ipv4.tcp_tso_win_divisor = 3
net.ipv4.tcp_congestion_control = cubic
net.ipv4.tcp_abc = 0
net.ipv4.tcp_mtu_probing = 0
net.ipv4.tcp_base_mss = 512
net.ipv4.tcp_workaround_signed_windows = 0
net.ipv4.tcp_slow_start_after_idle = 1
net.ipv4.tcp_available_congestion_control = cubic reno bic westwood highspeed hybla htcp vegas veno scalable lp yeah illinois
net.ipv4.tcp_allowed_congestion_control = cubic reno
net.ipv4.tcp_max_ssthresh = 0
net.ipv4.tcp_cookie_size = 0

-- 
Simon Arlott

[-- Attachment #2: c0.org.afilias-nst.info_ack_loop2.pcap.bz2 --]
[-- Type: application/x-bzip, Size: 64165 bytes --]

Re: TCP ACK loop with 2.6.33-rc6

[Ilpo Järvinen]

On Sat, 27 Feb 2010, Simon Arlott wrote:

> Packet capture attached. Is it possible to tell which side is at fault?
> They are both responding to each other's ACK with another ACK at 400pps.
> 
> 10:32:28.991111 IP6 2001:8b0:ffea:0:205:b4ff:fe12:530.35545 > 2001:500:b::1.53: . ack 766 win 58 <nop,nop,timestamp 2411331059 251057709>
> 10:32:28.995046 IP6 2001:500:b::1.53 > 2001:8b0:ffea:0:205:b4ff:fe12:530.35545: . ack 1 win 1559

Why only a single ends talks with timestamps?!?!?

-- 
 i.