-
Notifications
You must be signed in to change notification settings - Fork 193
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OVN: The ovn loadbalancers (ClusterIP) seem to choke on jumboframes / icmp requesting frag to the service #1022
Labels
Milestone
Comments
mangelajo
added
bug
Something isn't working
datapath
Datapath related issues or enhancements
OVN
labels
Dec 11, 2020
@mangelajo can you try if the following solution works with OVN? |
I tried, it doesn't work in this case with OVN. |
I will be using this instead (note the -d) to be more selective on what's exactly clamped
|
mangelajo
added a commit
to mangelajo/submariner
that referenced
this issue
Dec 14, 2020
OVN load balancing does not propagate fragment ICMPs down to the services, so when fragmentation becomes necessary the TCP stack of the service pod won't be able to detect this situation. Until this is fixed in OVN this workaround (to be enhanced later) will fix TCP. Fixes-Issue: submariner-io#1022 Signed-off-by: Miguel Angel Ajo <[email protected]>
mangelajo
added a commit
to mangelajo/submariner
that referenced
this issue
Dec 14, 2020
OVN load balancing does not propagate fragment ICMPs down to the services, so when fragmentation becomes necessary the TCP stack of the service pod won't be able to detect this situation. Until this is fixed in OVN this workaround (to be enhanced later) will fix TCP. Fixes-Issue: submariner-io#1022 Signed-off-by: Miguel Angel Ajo <[email protected]>
mangelajo
added a commit
to mangelajo/submariner
that referenced
this issue
Dec 14, 2020
OVN load balancing does not propagate fragment ICMPs down to the services, so when fragmentation becomes necessary the TCP stack of the service pod won't be able to detect this situation. Until this is fixed in OVN this workaround (to be enhanced later) will fix TCP. Fixes-Issue: submariner-io#1022 Signed-off-by: Miguel Angel Ajo <[email protected]>
tpantelis
pushed a commit
that referenced
this issue
Dec 14, 2020
OVN load balancing does not propagate fragment ICMPs down to the services, so when fragmentation becomes necessary the TCP stack of the service pod won't be able to detect this situation. Until this is fixed in OVN this workaround (to be enhanced later) will fix TCP. Fixes-Issue: #1022 Signed-off-by: Miguel Angel Ajo <[email protected]>
The MSS clamping workaround is in place, so we can close this one for now, although we must keep an eye on OVN for a proper fix. |
This was referenced Apr 21, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
What happened:
Our E2E tests fail since we test jumboframe fragmentation. (by chance... not by design)
https://mail.openvswitch.org/pipermail/ovs-discuss/2020-December/050834.html
iptables -I FORWARD -o ovn-k8s-gw0 -p tcp --tcp-flags SYN,RST SYN -j TCPMSS --set-mss 1398
What you expected to happen:
Service tests to work
How to reproduce it (as minimally and precisely as possible):
Anything else we need to know?:
Something that seems to workaround this, although it's not perfect, is to find the PMTU to the destination cluster....
And then clamp TCP/MSS with pmtu-40 (40 = IP + TCP overhead)
This won't help with UDP, but will help until we can figure out a solution in ovn. (if it's really in ovn.)
This was tried without success:
Environment:
subctl version
):kubectl version
):cat /etc/os-release
):uname -a
):The text was updated successfully, but these errors were encountered: