With continuous TCP session setup/destroy over a long time, few flows stuck in Hold state
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.20 |
Won't Fix
|
Medium
|
Praveen | |||
R3.0 |
Fix Committed
|
Medium
|
Praveen | |||
Trunk |
Fix Committed
|
Medium
|
Praveen |
Bug Description
R2.20 Build 112 Ubuntu 14.04 Juno multi-node setup
Between two VMs in a VN, below hping3 was creating 1000 tcp sessions and destroying them
for i in {1..1000000}; do hping3 -S -p 2200 219.14.75.3 -s 10000 -c 1000 -i u10000 ; done
Over a period of time, it was seen that few flows are stuck in hold state
After stopping all traffic, after over a min or so, these hold entries go away.
Gcore of the agent will be in http://
root@nodek1:~# flow -l | grep -B 1 Action:H
438268 219.14.75.4:10174 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:1070/57780, SPort:0)
--
924808 219.14.75.4:10664 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:4097/221238, SPort:0)
--
1145528 219.14.75.4:10777 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:2296/123984, SPort:0)
--
2520208 219.14.75.4:10685 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:665/35910, SPort:0)
--
3395648 219.14.75.4:10844 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:4831/260874, SPort:0)
--
3531352 219.14.75.4:10037 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:52/2808, SPort:0)
--
3933772 219.14.75.4:10591 219.14.75.3:2200 6 (2)
(K(nh):47, Action:H, Flags:, TCP:S, S(nh):0, Stats:4076/220104, SPort:0)
root@nodek1:~#
Changed in juniperopenstack: | |
importance: | High → Medium |
Review in progress for https:/ /review. opencontrail. org/17439
Submitter: Praveen K V (<email address hidden>)