Fix flaky e2e test TestL7NetworkPolicy #4723
Conversation
hongliangl
force-pushed
the
20230317-fix-4718-2
branch
from
f7d9e23 to
fd52765
Compare
hongliangl
changed the title
hongliangl
force-pushed
the
20230317-fix-4718-2
branch
2 times, most recently
from
502161f to
864b9f0
Compare
pkg/ovs/ovsctl/ovsctl.go
Outdated
| @@ -398,9 +398,15 @@ func (c *ovsCtlClient) DumpPortsDesc() ([][]string, error) { | |||
|
|
|||
| func (c *ovsCtlClient) SetPortNoFlood(ofport int) error { | |||
| // This command does not have standard output, and only has standard err when running with error. | |||
| _, err := c.ovsOfctlRunner.RunOfctlCmd("mod-port", strconv.FormatUint(uint64(ofport), 10), "no-flood") | |||
| // Note that, THIS CONFIGURATION MUST WORK WITH OpenFlow10. | |||
There was a problem hiding this comment.
🚀 just a question, looking at the changes in PR4517, could the bug also be caused by not capturing the standard err? because otherwise before the change we could use c.runOfctlCmd(false, "mod-port", ofport) to avoid Openflow15.
There was a problem hiding this comment.
just a question, looking at the changes in PR4517, could the bug also be caused by not capturing the standard err?
Impressive observation. However, I found that after executing the command 'ovs-ofctl mod-port br-int 6 no-flood -O Openflow15', there was no 'stout', and the 'no-flood' option was not configured.
because otherwise before the change we could use c.runOfctlCmd(false, "mod-port", ofport) to avoid Openflow15.
Thanks for reminding, we can use runOfctlCmd here.
hongliangl
force-pushed
the
20230317-fix-4718-2
branch
from
864b9f0 to
858c5e8
Compare
|
/skip-conformance |
Fixes antrea-io#4718 Signed-off-by: Hongliang Liu <lhongliang@vmware.com>
hongliangl
force-pushed
the
20230317-fix-4718-2
branch
from
858c5e8 to
91a3930
Compare
|
/test-e2e |
|
/skip-conformance |
Fixes antrea-io#4718 Signed-off-by: Hongliang Liu <lhongliang@vmware.com>
Fixes #4718
Signed-off-by: Hongliang Liu lhongliang@vmware.com