datapath: Do not SNAT replies to outside#17168
Merged
Conversation
49b6f6b to
6107b87
Compare
Member
Author
|
test-net-next |
76ad256 to
b9a4d2d
Compare
Member
Author
|
test-me-please |
Member
Author
|
test-1.19-5.4 |
Member
Author
|
test-1.20-4.19 |
1 similar comment
Member
Author
|
test-1.20-4.19 |
Member
Author
|
4.19 is hitting the complexity issue 👀 |
Member
Author
|
test-1.20-4.19 |
Member
Author
|
test-me-please |
12656d7 to
c6764f1
Compare
Member
Author
|
test-me-please |
c6764f1 to
b24f05b
Compare
Member
Author
|
test-me-please |
b24f05b to
d567075
Compare
Previously, the BPF-based masquerading (--enable-bpf-masquerade=true)
was wrongly masquerading replies from a pod to an outside when the
outside had initiated a connection. This was possible when e.g., the
outside had a route to the pod cidr.
To fix this, we introduce a lightweight CT lookup function
ct_is_reply4() which checks whether a given flow is a reply. The lookup
function is called in snat_v4_needed().
As a side note, I've tried to move the port extraction to a separate
function, but unfortunately it hits complexity issues on the 4.19
kernel in the "K8sDatapathConfig AutoDirectNodeRoutes Check direct
connectivity with per endpoint routes" suite:
BPF program is too large. Processed 131073 insn
libbpf: failed to load program 'handle_to_container'
libbpf: failed to load object '624_next/bpf_lxc.o'
Signed-off-by: Martynas Pumputis <m@lambda.lt>
Signed-off-by: Martynas Pumputis <m@lambda.lt>
Previously, they were failing due to our datapath masquerading replies from pod to outside. As it got fixed in the previous commit, we can enable BPF-based masquerading. This will also gives us some coverage for the fix. Signed-off-by: Martynas Pumputis <m@lambda.lt>
Member
Author
|
test-me-please Job 'Cilium-PR-K8s-1.16-net-next' failed and has not been observed before, so may be related to your PR: Click to show.Test NameFailure OutputIf it is a flake, comment |
Member
Author
|
test-net-next |
Member
|
CI has passed and we have reviewer approval, marking ready to merge. |
2 tasks
brb
added a commit
that referenced
this pull request
Jul 11, 2024
Previously, we required socketLB to be enabled in order for BPF
masquerade to properly function. The reasoning was outlined in [1] and
[2].
As pointed by Julian Wiedmann, [3] resolved the following NAT reply
issue:
On the remote node, the reply (dst=the client node IP) gets
masqueraded by the BPF-masq feature, because we masquerade pod ->
remote host IP in the tunnel mode (see comment in the
"snat_v4_needed()" for the reason), and currently we don't consult
the CT map to see whether a packet is reply.
Thus, we can remove the check.
[1]: #15437
[2]: 50e59c3
[3]: #17168
Signed-off-by: Martynas Pumputis <m@lambda.lt>
brb
added a commit
that referenced
this pull request
Jul 11, 2024
Previously, we required socketLB to be enabled in order for BPF
masquerade to properly function. The reasoning was outlined in [1] and
[2].
As pointed by Julian Wiedmann, [3] resolved the following NAT reply
issue:
On the remote node, the reply (dst=the client node IP) gets
masqueraded by the BPF-masq feature, because we masquerade pod ->
remote host IP in the tunnel mode (see comment in the
"snat_v4_needed()" for the reason), and currently we don't consult
the CT map to see whether a packet is reply.
Thus, we can remove the check.
[1]: #15437
[2]: 50e59c3
[3]: #17168
Signed-off-by: Martynas Pumputis <m@lambda.lt>
github-merge-queue Bot
pushed a commit
that referenced
this pull request
Jul 15, 2024
Previously, we required socketLB to be enabled in order for BPF
masquerade to properly function. The reasoning was outlined in [1] and
[2].
As pointed by Julian Wiedmann, [3] resolved the following NAT reply
issue:
On the remote node, the reply (dst=the client node IP) gets
masqueraded by the BPF-masq feature, because we masquerade pod ->
remote host IP in the tunnel mode (see comment in the
"snat_v4_needed()" for the reason), and currently we don't consult
the CT map to see whether a packet is reply.
Thus, we can remove the check.
[1]: #15437
[2]: 50e59c3
[3]: #17168
Signed-off-by: Martynas Pumputis <m@lambda.lt>
sayboras
pushed a commit
that referenced
this pull request
Jul 16, 2024
[ upstream commit 3de8537 ] Previously, we required socketLB to be enabled in order for BPF masquerade to properly function. The reasoning was outlined in [1] and [2]. As pointed by Julian Wiedmann, [3] resolved the following NAT reply issue: On the remote node, the reply (dst=the client node IP) gets masqueraded by the BPF-masq feature, because we masquerade pod -> remote host IP in the tunnel mode (see comment in the "snat_v4_needed()" for the reason), and currently we don't consult the CT map to see whether a packet is reply. Thus, we can remove the check. [1]: #15437 [2]: 50e59c3 [3]: #17168 Signed-off-by: Martynas Pumputis <m@lambda.lt> Signed-off-by: Tam Mach <tam.mach@cilium.io>
github-merge-queue Bot
pushed a commit
that referenced
this pull request
Jul 16, 2024
[ upstream commit 3de8537 ] Previously, we required socketLB to be enabled in order for BPF masquerade to properly function. The reasoning was outlined in [1] and [2]. As pointed by Julian Wiedmann, [3] resolved the following NAT reply issue: On the remote node, the reply (dst=the client node IP) gets masqueraded by the BPF-masq feature, because we masquerade pod -> remote host IP in the tunnel mode (see comment in the "snat_v4_needed()" for the reason), and currently we don't consult the CT map to see whether a packet is reply. Thus, we can remove the check. [1]: #15437 [2]: 50e59c3 [3]: #17168 Signed-off-by: Martynas Pumputis <m@lambda.lt> Signed-off-by: Tam Mach <tam.mach@cilium.io>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
8 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
See commit msgs.
Fix: #12544