TCP session teardown time in flows can be made to reflect the real teardown time
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.20 |
Fix Committed
|
Medium
|
Vedamurthy Joshi | |||
R3.0 |
Fix Committed
|
Medium
|
Vedamurthy Joshi | |||
Trunk |
Fix Committed
|
Medium
|
Vedamurthy Joshi |
Bug Description
R2.20 Build 115 Ubuntu 14.04 Juno multinode setup
From VM 219.14.75.4 , i ran hping3 to 219.14.75.3
root@vm2:
HPING 219.14.75.3 (eth0 219.14.75.3): S set, 40 headers + 0 data bytes
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=0 win=0 rtt=2.9 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=1 win=0 rtt=1.8 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=2 win=0 rtt=2.0 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=3 win=0 rtt=2.9 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=4 win=0 rtt=2.0 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=5 win=0 rtt=1.8 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=6 win=0 rtt=1.8 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=7 win=0 rtt=1.7 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=8 win=0 rtt=1.5 ms
len=40 ip=219.14.75.3 ttl=64 DF id=0 sport=33000 flags=RA seq=9 win=0 rtt=1.6 ms
--- 219.14.75.3 hping statistic ---
10 packets transmitted, 10 packets received, 0% packet loss
round-trip min/avg/max = 1.5/2.0/2.9 ms
root@vm2:
=======
So, each TCP session lasted for less than a second.
The flows get created, but the teardown time is shown as empty for these flows until the flow ageout time(180 s here)
After that, the teardown time is updated so that (teardown_time - setup_time) is greater than 180s
Which is not true. Since vrouter can figure out the real teardown time within the flow ageout window, we could reflect that in the flow stats
Attached are the flow stats in this example
bug update...