[Bug 252913] [tcp] Using RACK leaks mbufs

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=252913

--- Comment #29 from iron.udjin at gmail.com ---
After 10 hours uptime I have 3 looping connection which produce 30Mibt/s of
traffic. I cannot say exact time when they started looping because traffic
graph has quite linear grow: https://prnt.sc/xv6ldt

netstat -m:

265055/26920/291975 mbufs in use (current/cache/total)
130635K/124086K/254721K bytes allocated to network (current/cache/total)

# sysctl net.inet.tcp.functions_available net.inet.tcp.rack
net.inet.tcp.functions_available:
Stack                           D Alias                            PCB count
freebsd                           freebsd                          3
rack                            * rack                             4107

net.inet.tcp.rack.clear: 0
net.inet.tcp.rack.tlp_nada: 0
net.inet.tcp.rack.idle_reduce_oninput: 0
net.inet.tcp.rack.prog_drops: 0
net.inet.tcp.rack.runtsacks: 854985
net.inet.tcp.rack.hit_tlp_method2: 43498272
net.inet.tcp.rack.hit_tlp_method: 26318731
net.inet.tcp.rack.tlp_calc_entered: 143382781
net.inet.tcp.rack.sack_short: 4099975
net.inet.tcp.rack.sack_restart: 0
net.inet.tcp.rack.sack_long: 466643
net.inet.tcp.rack.split_limited: 7165
net.inet.tcp.rack.alloc_limited_conns: 45
net.inet.tcp.rack.alloc_limited: 51035137
net.inet.tcp.rack.allocemerg: 0
net.inet.tcp.rack.allochard: 0
net.inet.tcp.rack.allocs: 107989742
net.inet.tcp.rack.saw_enetunreach: 0
net.inet.tcp.rack.saw_enobufs: 0
net.inet.tcp.rack.unpaced: 138312500
net.inet.tcp.rack.paced: 9822106
net.inet.tcp.rack.arm_tlp: 0
net.inet.tcp.rack.arm_rack: 0
net.inet.tcp.rack.rack_to_tot: 51807262
net.inet.tcp.rack.tlp_retran_fail: 43664
net.inet.tcp.rack.tlp_retran_bytes: 249582917
net.inet.tcp.rack.tlp_retran: 487864
net.inet.tcp.rack.tlp_new: 80270
net.inet.tcp.rack.tlp_to_total: 543162
net.inet.tcp.rack.reordering: 686406
net.inet.tcp.rack.findhigh: 26081
net.inet.tcp.rack.tsnf: 0
net.inet.tcp.rack.prrsndnew: 2614014
net.inet.tcp.rack.prrsndret: 55895077
net.inet.tcp.rack.badfr_bytes: 255611021
net.inet.tcp.rack.badfr: 241600
net.inet.tcp.rack.inc_var: 0
net.inet.tcp.rack.pktdelay: 1
net.inet.tcp.rack.reorder_fade: 60000
net.inet.tcp.rack.rtt_tlp_thresh: 1
net.inet.tcp.rack.reorder_thresh: 2
net.inet.tcp.rack.earlyrecovery: 1
net.inet.tcp.rack.earlyrecoveryseg: 6
net.inet.tcp.rack.minto: 1
net.inet.tcp.rack.prr_sendalot: 1
net.inet.tcp.rack.hptsi_seg_max: 40
net.inet.tcp.rack.hptsi_every_seg: 1
net.inet.tcp.rack.hptsi_reduces: 4
net.inet.tcp.rack.tlp_cwnd_flag: 0
net.inet.tcp.rack.recovery_prop: 10
net.inet.tcp.rack.recovery_loss_prop: 0
net.inet.tcp.rack.tlp_retry: 2
net.inet.tcp.rack.maxrto: 30000
net.inet.tcp.rack.minrto: 30
net.inet.tcp.rack.rack_tlimit: 0
net.inet.tcp.rack.rack_tlp_in_recovery: 1
net.inet.tcp.rack.send_oldest: 0
net.inet.tcp.rack.sblklimit: 128
net.inet.tcp.rack.precache: 1
net.inet.tcp.rack.tlpminto: 10
net.inet.tcp.rack.delayed_ack: 200
net.inet.tcp.rack.sackfiltering: 1
net.inet.tcp.rack.bb_verbose: 0
net.inet.tcp.rack.idle_reduce_high: 0
net.inet.tcp.rack.min_pace_segs: 6
net.inet.tcp.rack.min_pace_time: 0
net.inet.tcp.rack.tlpmethod: 2
net.inet.tcp.rack.data_after_close: 1
net.inet.tcp.rack.rate_sample_method: 1
net.inet.tcp.rack.map_splitlimit: 256
net.inet.tcp.rack.map_limit: 1024


# sockstat -sSPtcp | grep AAA.AAA.AAA.AAA
?        ?          ?     ?  tcp4   XXX.XXX.XXX.XXX:443     
AAA.AAA.AAA.AAA:62799                 LAST_ACK     rack
# sockstat -sSPtcp | grep BBB.BBB.BBB.BBB
?        ?          ?     ?  tcp4   XXX.XXX.XXX.XXX:443     
BBB.BBB.BBB.BBB:32440                 LAST_ACK     rack
# sockstat -sSPtcp | grep CCC.CCC.CCC.CCC
?        ?          ?     ?  tcp4   XXX.XXX.XXX.XXX:443     
CCC.CCC.CCC.CCC:50608                CLOSING      rack

-- 
You are receiving this mail because:
You are on the CC list for the bug.