This makes the "destination" policy model clearer. All external
destination captures now jump to the "XLB chain, which is the main place
that masquerade is done (removing it from most other places).
This is simpler to trace - XLB *always* exists (as long as you have an
external exposure) and never gets bypassed.
No functional changes, much whitespace.
Make assertIPTablesRulesEqual() *not* sort the `expected` value - make
the test cases all be pre-sorted. This will make followup commits
cleaner.
Make the test output cleaner when this fails.
Use dedent everywhere for easier reading.
Fix internal and external traffic policy to be handled separately (so
that, in particular, services with Local internal traffic policy and
Cluster external traffic policy do not behave as though they had Local
external traffic policy as well.
Additionally, traffic to an `internalTrafficPolicy: Local` service on
a node with no endpoints is now dropped rather than being rejected
(which, as in the external case, may prevent traffic from being lost
when endpoints are in flux).
Now the XLB chain _only_ implements the "short-circuit local
connections to the SVC chain" rule, and the actual endpoint selection
happens in the SVL chain.
Though not quite implemented yet, this will eventually also mean that
"SVC" = "Service, Cluster traffic policy" as opposed to "SVL" =
"Service, Local traffic policy"
Rather than lazily computing and then caching the endpoint chain name
because we don't have the right information at construct time, just
pass the right information at construct time and compute the chain
name then.
Now that we don't have to always append all of the iptables args into
a single array, there's no reason to have LocalTrafficDetector take in
a set of args to prepend to its own output, and also not much point in
having it write out the "-j CHAIN" by itself either.
For each iptables-restore call, log the number of services, endpoints,
filter chains, filter rules, NAT chains, and NAT rules in the update
at V(2), in addition to logging the actual rules if V(9).
All of the tests used a localDetector that considered the pod IP range
to be 10.0.0.0/24, but lots of the tests used pod IPs in 10.180.0.0/16
or 10.0.1.0/24, meaning the generated iptables rules were somewhat
inconsistent. Fix this by expanding the localDetector's pod IP range
to 10.0.0.0/8. (Changing the pod IPs to all be in 10.0.0.0/24 instead
would be a much larger change since it would result in the SEP chain
names changing.)
Meanwhile, the different tests were also horribly inconsistent about
what values they used for other IPs, and some of them even used the
same IPs (or ports) for different things in the same test case. Fix
these all up and create a consistent set of IP assignments:
// Pod IPs: 10.0.0.0/8
// Service ClusterIPs: 172.30.0.0/16
// Node IPs: 192.168.0.0/24
// Local Node IP: 192.168.0.2
// Service ExternalIPs: 192.168.99.0/24
// LoadBalancer IPs: 1.2.3.4, 5.6.7.8, 9.10.11.12
// Non-cluster IPs: 203.0.113.0/24
// LB Source Range: 203.0.113.0/25
Only run assertIPTablesRuleJumps() on the expected output, not on the
actual output, since if there's a problem with the actual output, we'd
rather see it as the diff from the expected output.
In large clusters, the iptables-restore input will be tens of
thousands of lines long, and logging it at V(5) essentially means that
"kube-proxy -v=5" cannot be used in such clusters to see _other_
things that get logged at V(5), because logs will get rolled over far
too quickly. So bump the full-rules logging output down to V(9).
kube-proxy sets the sysctl net.ipv4.conf.all.route_localnet=1
so NodePort services can be accessed on the loopback addresses in
IPv4, but this may present security issues.
Leverage the --nodeport-addresses flag to opt-out of this feature,
if the list is not empty and none of the IP ranges contains an IPv4
loopback address this sysctl is not set.
In addition, add a warning to inform users about this behavior.
When nodePortAddresses is not specified for kube-proxy, it tried to open
the node port for a NodePort service twice, triggered by IPv4ZeroCIDR
and IPv6ZeroCIDR separately. The first attempt would succeed and the
second one would always generate an error log like below:
"listen tcp4 :30522: bind: address already in use"
This patch fixes it by ensuring nodeAddresses of a proxier only contain
the addresses for its IP family.
The same code appeared twice, once for the SVC chain and once for the
XLB chain, with the only difference being that the XLB version had
more verbose comments.
Also, in the NodePort code, fix it to properly take advantage of the
fact that GetNodeAddresses() guarantees that if it returns a
"match-all" CIDR, then it doesn't return anything else. That also
makes it unnecessary to loop over the node addresses twice.
If you pass just an IP address to "-s" or "-d", the iptables command
will fill in the correct mask automatically.
Originally, the proxier was just hardcoding "/32" for all of these,
which was unnecessary but simple. But when IPv6 support was added, the
code was made more complicated to deal with the fact that the "/32"
needed to be "/128" in the IPv6 case, so it would parse the IPs to
figure out which family they were, which in turn involved adding some
checks in case the parsing fails (even though that "can't happen" and
the old code didn't check for invalid IPs, even though that would
break the iptables-restore if there had been any).
Anyway, all of that is unnecessary because we can just pass the IP
strings to iptables directly rather than parsing and unparsing them
first.
(The diff to proxier_test.go is just deleting "/32" everywhere.)
If GetNodeAddresses() fails (eg, because you passed the wrong CIDR to
`--nodeport-addresses`), then any NodePort services would end up with
only half a set of iptables rules. Fix it to just not output the
NodePort-specific parts in that case (in addition to logging an error
about the GetNodeAddresses() failure).
The iptables and ipvs proxiers both had a check that none of the
elements of svcInfo.LoadBalancerIPStrings() were "", but that was
already guaranteed by the svcInfo code. Drop the unnecessary checks
and remove a level of indentation.
* Fix regression in kube-proxy
Don't use a prepend() - that allocates. Instead, make Write() take
either strings or slices (I wish we could express that better).
* WIP: switch to intf
* WIP: less appends
* tests and ipvs
The logic to detect stale endpoints was not assuming the endpoint
readiness.
We can have stale entries on UDP services for 2 reasons:
- an endpoint was receiving traffic and is removed or replaced
- a service was receiving traffic but not forwarding it, and starts
to forward it.
Add an e2e test to cover the regression
Filter the allEndpoints list into readyEndpoints sooner, and set
"hasEndpoints" based (mostly) on readyEndpoints, not allEndpoints (so
that, eg, we correctly generate REJECT rules for services with no
_functioning_ endpoints, even if they have unusable terminating
endpoints).
Also, write out the endpoint chains at the top of the loop when we
iterate the endpoints for the first time, rather than copying some of
the data to another set of variables and then writing them out later.
And don't write out endpoint chains that won't be used
Also, generate affinity rules only for readyEndpoints rather than
allEndpoints, so affinity gets broken correctly when an endpoint
becomes unready.
The external traffic policy terminating endpoints test was testing
LoadBalancer functionality against a NodePort service with no
nodePorts (or loadBalancer IPs). It managed to test what it wanted to
test, but it's kind of dubious (and we probably _shouldn't_ have been
generating the rules it was looking for since there was no way to
actually reach the XLB chains). So fix that.
Also make the terminating endpoints test use session affinity, to add
more testing for that. Also, remove the multiple copies of the same
identical Service that is used for all of the test cases in that test.
Also add a "Cluster traffic policy and no source ranges" test to
TestOverallIPTablesRulesWithMultipleServices since we weren't really
testing either of those.
Also add a test of --masquerade-all.
The test got broken to not actually use "no cluster CIDR" when
LocalDetector was implemented (and the old version of the unit test
didn't check enough to actually notice this).
The original tests here were very shy about looking at the iptables
output, and just relied on checks like "make sure there's a jump to
table X that also includes string Y somewhere in it" and stuff like
that. Whereas the newer tests were just like, "eh, here's a wall of
text, make sure the iptables output is exactly that". Although the
latter looks messier in the code, it's more precise, and it's easier
to update correctly when you change the rules. So just make all of the
tests do a check on the full iptables output.
(Note that I didn't double-check any of the output; I'm just assuming
that the output of the current iptables proxy code is actually
correct...)
Also, don't hardcode the expected number of rules in the metrics
tests, so that there's one less thing to adjust when rules change.
Also, use t.Run() in one place to get more precise errors on failure.
The test was sorting the iptables output so as to not depend on the
order that services get processed in, but this meant it wasn't
checking the relative ordering of rules (and in fact, the ordering of
the rules in the "expected" string was wrong, in a way that would
break things if the rules had actually been generated in that order).
Add a more complicated sorting function that sorts services
alphabetically while preserving the ordering of rules within each
service.
Because the proxy.Provider interface included
proxyconfig.EndpointsHandler, all the backends needed to
implement its methods. But iptables, ipvs, and winkernel implemented
them as no-ops, and metaproxier had an implementation that wouldn't
actually work (because it couldn't handle Services with no active
Endpoints).
Since Endpoints processing in kube-proxy is deprecated (and can't be
re-enabled unless you're using a backend that doesn't support
EndpointSlice), remove proxyconfig.EndpointsHandler from the
definition of proxy.Provider and drop all the useless implementations.
The nat KUBE-SERVICES chain is called from OUTPUT and PREROUTING stages. In
clusters with large number of services, the nat-KUBE-SERVICES chain is the largest
chain with for eg: 33k rules. This patch aims to move the KubeMarkMasq rules from
the kubeServicesChain into the respective KUBE-SVC-* chains. This way during each
packet-rule matching we won't have to traverse the MASQ rules of all services which
get accumulated in the KUBE-SERVICES and/or KUBE-NODEPORTS chains. Since the
jump to KUBE-MARK-MASQ ultimately sets the 0x400 mark for nodeIP SNAT, it should not
matter whether the jump is made from KUBE-SERVICES or KUBE-SVC-* chains.
Specifically we change:
1) For ClusterIP svc, we move the KUBE-MARK-MASQ jump rule from KUBE-SERVICES
chain into KUBE-SVC-* chain.
2) For ExternalIP svc, we move the KUBE-MARK-MASQ jump rule in the case of
non-ServiceExternalTrafficPolicyTypeLocal from KUBE-SERVICES
chain into KUBE-SVC-* chain.
3) For NodePorts svc, we move the KUBE-MARK-MASQ jump rule in case of
non-ServiceExternalTrafficPolicyTypeLocal from KUBE-NODEPORTS chain to
KUBE-SVC-* chain.
4) For load-balancer svc, we don't change anything since it is already svc specific
due to creation of KUBE-FW-* chains per svc.
This would cut the rules per svc in KUBE-SERVICES and KUBE-NODEPORTS in half.
1. Do not describe port type in message because lp.String() already has the
information.
2. Remove duplicate error detail from event log.
Previous log is like this.
47s Warning listen tcp4 :30764: socket: too many open files node/127.0.0.1 can't open port "nodePort for default/temp-svc:834" (:30764/tcp4), skipping it: listen tcp4 :30764: socket: too many open files
[issue]
When creating a NodePort service with the kubectl create command, the NodePort
assignment may fail.
Failure to assign a NodePort can be simulated with the following malicious
command[1].
$ kubectl create service nodeport temp-svc --tcp=`python3 <<EOF
print("1", end="")
for i in range(2, 1026):
print("," + str(i), end="")
EOF
`
The command succeeds and shows following output.
service/temp-svc created
The service has been successfully generated and can also be referenced with the
get command.
$ kubectl get svc
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S)
temp-svc NodePort 10.0.0.139 <none> 1:31335/TCP,2:32367/TCP,3:30263/TCP,(omitted),1023:31821/TCP,1024:32475/TCP,1025:30311/TCP 12s
The user does not recognize failure to assign a NodePort because
create/get/describe command does not show any error. This is the issue.
[solution]
Users can notice errors by looking at the kube-proxy logs, but it may be difficult to see the kube-proxy logs of all nodes.
E0327 08:50:10.216571 660960 proxier.go:1286] "can't open port, skipping this nodePort" err="listen tcp4 :30641: socket: too many open files" port="\"nodePort for default/temp-svc:744\" (:30641/tcp4)"
E0327 08:50:10.216611 660960 proxier.go:1286] "can't open port, skipping this nodePort" err="listen tcp4 :30827: socket: too many open files" port="\"nodePort for default/temp-svc:857\" (:30827/tcp4)"
...
E0327 08:50:10.217119 660960 proxier.go:1286] "can't open port, skipping this nodePort" err="listen tcp4 :32484: socket: too many open files" port="\"nodePort for default/temp-svc:805\" (:32484/tcp4)"
E0327 08:50:10.217293 660960 proxier.go:1612] "Failed to execute iptables-restore" err="pipe2: too many open files ()"
I0327 08:50:10.217341 660960 proxier.go:1615] "Closing local ports after iptables-restore failure"
So, this patch will fire an event when NodePort assignment fails.
In fact, when the externalIP assignment fails, it is also notified by event.
The event will be displayed like this.
$ kubectl get event
LAST SEEN TYPE REASON OBJECT MESSAGE
...
2s Warning listen tcp4 :31055: socket: too many open files node/127.0.0.1 can't open "nodePort for default/temp-svc:901" (:31055/tcp4), skipping this nodePort: listen tcp4 :31055: socket: too many open files
2s Warning listen tcp4 :31422: socket: too many open files node/127.0.0.1 can't open "nodePort for default/temp-svc:474" (:31422/tcp4), skipping this nodePort: listen tcp4 :31422: socket: too many open files
...
This PR fixes iptables and ipvs proxier.
Since userspace proxier does not seem to be affected by this issue, it is not fixed.
[1] Assume that fd limit is 1024(default).
$ ulimit -n
1024
1. Add API definitions;
2. Add feature gate and drops the field when feature gate is not on;
3. Set default values for the field;
4. Add API Validation
5. add kube-proxy iptables and ipvs implementations
6. add tests
Clear conntrack entries for UDP NodePorts,
this has to be done AFTER the iptables rules are programmed.
It can happen that traffic to the NodePort hits the host before
the iptables rules are programmed this will create an stale entry
in conntrack that will blackhole the traffic, so we need to
clear it ONLY when the service has endpoints.
1. For iptables mode, add KUBE-NODEPORTS chain in filter table. Add
rules to allow healthcheck node port traffic.
2. For ipvs mode, add KUBE-NODE-PORT chain in filter table. Add
KUBE-HEALTH-CHECK-NODE-PORT ipset to allow traffic to healthcheck
node port.
Currently kube-proxy treat ExternalIPs differently depending on:
- the traffic origin
- if the ExternalIP is present or not in the system.
It also depends on the CNI implementation to
discriminate between local and non-local traffic.
Since the ExternalIP belongs to a Service, we can avoid the roundtrip
of sending outside the traffic originated in the cluster.
Also, we leverage the new LocalTrafficDetector to detect the local
traffic and not rely on the CNI implementations for this.
