14.1X53D26/Contrail: after restart ovsdb-server it takes more than 15+ min re-sync config / recover the traffic
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Juniper Openstack | Status tracked in Trunk | |||||
R2.20 |
Fix Committed
|
Medium
|
Hari Prasad Killi | |||
Trunk |
Fix Committed
|
Medium
|
Hari Prasad Killi |
Bug Description
Hi Rahul,
Sudarsan just debugged this issue, and confirmed that its contrail issue..
Whenever we do "restart ovsdb-server" contrail is deleting and adding remote mac, that's why we are seeing traffic drop on the remote node
I will update the PR with the logs, please assign the PR accordingly
Thanks,
Praveen
vi /var/db/
OVSDB JSON 90 837ead3587d0db6
{"Ucast_
OVSDB JSON 90 fe579f67deb6675
{"Ucast_
OVSDB JSON 90 3c3224db3c2d980
{"Ucast_
OVSDB JSON 90 fc86b598c88490a
{"Ucast_
OVSDB JSON 90 449c913b68cfb88
{"Ucast_
OVSDB JSON 236 9961aac995a4e1f
{"Ucast_
937a-ebd6-
OVSDB JSON 236 150d9b675bfc2c5
{"Ucast_
937a-ebd6-
OVSDB JSON 236 453b1346eb0ad0a
{"Ucast_
937a-ebd6-
OVSDB JSON 236 b7ffaa212551b7b
{"Ucast_
937a-ebd6-
OVSDB JSON 236 fd73e08756673e6
{"Ucast_
937a-ebd6-
OVSDB JSON 236 bba2a93abd9d399
{"Ucast_
-------
after restart ovsdb-server it takes more than 15+ min re-sync config / recover the traffic
Traffic profile:
bi-dir L2 unicast traffic between DUT and 8 hw vteps at the rate of 1000pps
root@st-24q-p2-8> show ethernet-switching table summary
Total dynamic and static MAC addresses learned globally : 22776
Configured static MAC addresses learned globally : 14856
config:
4K LIFs
1K VNI
1k VTEP
have copied the gcore and ovsdb logs here
[pvshetty@
/volume/
[pvshetty@
ovs_core ovsdb.log.0.gz ovsdb.log.12.gz ovsdb.log.15.gz ovsdb.log.18.gz ovsdb.log.3.gz ovsdb.log.6.gz ovsdb.log.9.gz
ovs_core-1 ovsdb.log.10.gz ovsdb.log.13.gz ovsdb.log.16.gz ovsdb.log.1.gz ovsdb.log.4.gz ovsdb.log.7.gz
ovsdb.log ovsdb.log.11.gz ovsdb.log.14.gz ovsdb.log.17.gz ovsdb.log.2.gz ovsdb.log.5.gz ovsdb.log.8.gz
tags: | added: vrouter |
Changed in juniperopenstack: | |
importance: | Undecided → Medium |
tags: | added: bms |
tags: | added: scale |
information type: | Proprietary → Public |
This was the behaviour in R2.1. An audit mechanism has been added as part of R2.2. Please recheck with an R2.2 build.