-
Notifications
You must be signed in to change notification settings - Fork 241
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
ci:[CNI] Add windows CNIv1 datapath test #2016
Conversation
eff2ce8
to
9709a1b
Compare
/azp run |
Azure Pipelines successfully started running 2 pipeline(s). |
t.Log("Get Nodes") | ||
nodes, err := k8sutils.GetNodeListByLabelSelector(ctx, clientset, nodeLabelSelector) | ||
if err != nil { | ||
require.NoError(t, err, "could not get k8s node list: %v", err) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why require.error
and t.fatal
at other places ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Left them in when testing and forgot to change. Will change to require.NoError for better reporting.
@@ -156,7 +169,7 @@ func MustCreateNamespace(ctx context.Context, clienset *kubernetes.Clientset, na | |||
}, | |||
}, metav1.CreateOptions{}) | |||
|
|||
if !apierrors.IsAlreadyExists(err) { | |||
if apierrors.IsAlreadyExists(err) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
So IsAlreadyExists(err)
will return true if err is related to already existing resource
.
This is bypassing all other errors. What we want :
- If there is
already existing resource
ornil
error then return nil. - Otherwise return the actual error.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Okay, should we rename the function from MustCreateNamespace -> NamespaceMustExist? To capture that we want to create the namespace or use the one that is existing?
!IsAlreadyExists(err)
will never catch anything as well. As err would have to be nil for it to report false and then line 173 would be a false positive.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can add additional error checking to capture errors other than metav1.StatusReasonAlreadyExists
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think that would be better to add additional error checking. Thanks !
t.Log("Successfully created customer windows pods") | ||
} else { | ||
// Checks namespace already exists from previous attempt | ||
t.Log("Namespace already exists") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we want to check if pods exists as well or not ?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We should. No reason not to.
return nil | ||
} | ||
|
||
func invokeWebRequestPassedWindows(output string) error { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is not invoking web request, this is checking the result. So we wan change the name of the function appropriately.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Changed.
c9d6a22
to
1478aa4
Compare
if err != nil { | ||
return errors.Wrap(err, fmt.Sprintf("Getting pod %s failed with %v", firstPod.Name, err)) | ||
} | ||
logrus.Infof("First pod: %v", firstPod.Name) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we also print this pod IP addresses here (ipv4 && ipv6) so that we can know source and destination IP that's more clear.
I see you print ip address on second pod below:
logrus.Infof("Second pod: %v %v", secondPod.Name, secondPod.Status.PodIP)
b8eb182
to
f567793
Compare
f380118
to
9fb78ba
Compare
9fb78ba
to
7875da5
Compare
spec: | ||
containers: | ||
- name: windows-container | ||
image: mcr.microsoft.com/dotnet/framework/samples:aspnetapp |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sorry I missed this in the initial review. Can we use a smaller image ? The size is way aroung 1.5 Gb.
This one: mcr.microsoft.com/windows/nanoserver:ltsc2022 size is around 100mb, but not sure if it resolves all the cmds you want to execute.
11cdc4f
to
761d570
Compare
761d570
to
3da7b46
Compare
* ci: Transfer files * test: Working Datapath Test * test: apierror Tests * style: Datapath Package * test: Deployment timing * fix: Error check * fix: Lint (cherry picked from commit 390977d)
* ci: Transfer files * test: Working Datapath Test * test: apierror Tests * style: Datapath Package * test: Deployment timing * fix: Error check * fix: Lint (cherry picked from commit 390977d)
* ci: Transfer files * test: Working Datapath Test * test: apierror Tests * style: Datapath Package * test: Deployment timing * fix: Error check * fix: Lint (cherry picked from commit 390977d)
* build azure-vnet-telemetry and azure-vnet-ipam in dropgz-test (#1846) build azure-vnet-telemetry and azure-vnet-ipam in dropgz-test for parity with release image Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> (cherry picked from commit f619259) * ci: disable kube-proxy for test clusters (#1965) * disable kube-proxy for byocni cluster creation * test config mapping * shell pwd * use CURDIR * check current directory * test with repo root dir * test azp format * test azp format * test azp format * change e2e steps to remove kube proxy * fix load test update args * fix ns and rg in update * update ciliume2e * fix kubectl cmd in load test * adding new targets for no kube proxy * remove cluster update * update overlay e2e * test behavior of load test * test grep for azure-cns * look for container deployment * testing * restart node variable check * update if condition * add skip node case --------- Co-authored-by: tamilmani1989 <tamanoha@microsoft.com> (cherry picked from commit 024819d) * CI: [CNI] Replace the bash scripts for CNI load testing with golang test cases (#2003) CI:[CNI] Replace the bash scripts with the golang test cases (cherry picked from commit 008ae45) * ci: [CNI] Move Nightly Cilium Pipeline test to ACN (#1963) * CNS to be able to generate dualstack overaly CNI conflist (#1981) * fix: Eliminating duplicate lines * ci: Add update permission for ciliumidentity * fix: Parameterize Image Registry add retry to nnc update during scaledown (#1970) * add retry to nnc update during scaledown Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> * test for panic in pool monitor Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> --------- Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> fix: reserve 0th IP as gateway for overlay on Windows (#1968) * fix: reserve 0th IP as gateway for overlay on Windows * fix: allow gateway to be updated ci: windows profile container image (#1988) Always use 0 for NC version in Overlay (#1979) always use 0 for NC version in overlay Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> [Vnet Scale - CNS]: Flattening CIDR ranges for Node NNC to a list (#1921) * Read secondary CIDRs from VnetScale NNC * fix comment * update comment * For VnetScale mode, Use 1st IP for def gateway instead of 0th for windows * fix/add import * address pr comments * add comments * address pr comments * wrap error * fix typo * fix UT fix: [NPM] check if policy exists in case of nil pointer (#1974) fix: check for nil first ci: disable kube-proxy for test clusters (#1965) * disable kube-proxy for byocni cluster creation * test config mapping * shell pwd * use CURDIR * check current directory * test with repo root dir * test azp format * test azp format * test azp format * change e2e steps to remove kube proxy * fix load test update args * fix ns and rg in update * update ciliume2e * fix kubectl cmd in load test * adding new targets for no kube proxy * remove cluster update * update overlay e2e * test behavior of load test * test grep for azure-cns * look for container deployment * testing * restart node variable check * update if condition * add skip node case --------- Co-authored-by: tamilmani1989 <tamanoha@microsoft.com> perf: [WIN-NPM] fast bootup (#1900) * wip * wip2 * use other apply DP func * address comment about if statement * finish bootup for both DPs * fix lint * fix lint 2 * fix lint 3 * longer UT timeout and add missing UTs for apply in background tool: [NPM] script to clean up iptable chains (#1978) tool: script to clean up NPM iptable chains feat: [WIN-NPM] metrics for latencies and failures (#1959) * implement metrics * add npm prefix * rename windows files * metrics pkg UTs * allow reinitializing prometheus metrics * fix: hns wrapper should not throw error for empty SetPolicy values * test: metric UTs in dataplane * fix: record list endpoint latency always * remove flaky UT * feat: metric for max ipset members * fix lint * fix lint 2 * fix build * fix lint 3 * simplify conditionals and protect against maxMembers becoming negative * remove bottom 4 histogram buckets. start at 16 ms * reset metrics for ipset UTs * style: don't check for windows dp in *_windows.go files * build: remove unused import * test: reset windows metrics in UT Remove SSH port 22 rule from aks-engine clusters (#1983) ci: change overlaye2e stage to cilium-overlay (#1997) * renaming overlaye2e for cilium * update display names for stages Initial getHomeAZ 404 changes (#1994) * initial getHomeAZ 404 changes * treat 404 as success * address comments CNS to be able to generate dualstack overaly CNI conflist (#1981) fix: Parameterize Image Registry add retry to nnc update during scaledown (#1970) * add retry to nnc update during scaledown Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> * test for panic in pool monitor Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> --------- Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> fix: reserve 0th IP as gateway for overlay on Windows (#1968) * fix: reserve 0th IP as gateway for overlay on Windows * fix: allow gateway to be updated ci: windows profile container image (#1988) Always use 0 for NC version in Overlay (#1979) always use 0 for NC version in overlay Signed-off-by: Evan Baker <rbtr@users.noreply.github.com> [Vnet Scale - CNS]: Flattening CIDR ranges for Node NNC to a list (#1921) * Read secondary CIDRs from VnetScale NNC * fix comment * update comment * For VnetScale mode, Use 1st IP for def gateway instead of 0th for windows * fix/add import * address pr comments * add comments * address pr comments * wrap error * fix typo * fix UT fix: [NPM] check if policy exists in case of nil pointer (#1974) fix: check for nil first ci: disable kube-proxy for test clusters (#1965) * disable kube-proxy for byocni cluster creation * test config mapping * shell pwd * use CURDIR * check current directory * test with repo root dir * test azp format * test azp format * test azp format * change e2e steps to remove kube proxy * fix load test update args * fix ns and rg in update * update ciliume2e * fix kubectl cmd in load test * adding new targets for no kube proxy * remove cluster update * update overlay e2e * test behavior of load test * test grep for azure-cns * look for container deployment * testing * restart node variable check * update if condition * add skip node case --------- Co-authored-by: tamilmani1989 <tamanoha@microsoft.com> perf: [WIN-NPM] fast bootup (#1900) * wip * wip2 * use other apply DP func * address comment about if statement * finish bootup for both DPs * fix lint * fix lint 2 * fix lint 3 * longer UT timeout and add missing UTs for apply in background tool: [NPM] script to clean up iptable chains (#1978) tool: script to clean up NPM iptable chains feat: [WIN-NPM] metrics for latencies and failures (#1959) * implement metrics * add npm prefix * rename windows files * metrics pkg UTs * allow reinitializing prometheus metrics * fix: hns wrapper should not throw error for empty SetPolicy values * test: metric UTs in dataplane * fix: record list endpoint latency always * remove flaky UT * feat: metric for max ipset members * fix lint * fix lint 2 * fix build * fix lint 3 * simplify conditionals and protect against maxMembers becoming negative * remove bottom 4 histogram buckets. start at 16 ms * reset metrics for ipset UTs * style: don't check for windows dp in *_windows.go files * build: remove unused import * test: reset windows metrics in UT Remove SSH port 22 rule from aks-engine clusters (#1983) ci: change overlaye2e stage to cilium-overlay (#1997) * renaming overlaye2e for cilium * update display names for stages Initial getHomeAZ 404 changes (#1994) * initial getHomeAZ 404 changes * treat 404 as success * address comments CNS to be able to generate dualstack overaly CNI conflist (#1981) * fix: File Directory * style: Comments * Addressing Comments --------- Co-authored-by: Paul Johnston <35265851+pjohnst5@users.noreply.github.com> (cherry picked from commit 1514d95) * ci:[CNI] Add windows CNIv1 datapath test (#2016) * ci: Transfer files * test: Working Datapath Test * test: apierror Tests * style: Datapath Package * test: Deployment timing * fix: Error check * fix: Lint (cherry picked from commit 390977d) * fix: [CNI] CNI load test failing due to namespace already created (#2031) fix: CNI load test failing due to namespace already created (cherry picked from commit c10900e) * ci:[CNI] Windows cniv1 load test pipeline (#2024) CI:[CNI] Windows cniv1 load test pipeline (cherry picked from commit e45ad21) * ci: [CNI] Adding aks cluster creation steps for k8s e2e test (#2052) * ci: [CNI] Adding aks cluster creation steps for k8s e2e test * Add validate step to the pipeline * Adding the telemetry config to the cluster (cherry picked from commit 846e508) * ci:[CNI] Replace AKS-Engine Tests with k8s conformance tests (#2062) * Initial Commit * Add attempts to prevent flakyness * Add taint for windows tests * Add k8s e2e tests * Testing vmSizes * Artifact k8se2e binary * Remove NPM E2E * Add testing and increase processes * Addressing comments (cherry picked from commit 451c691) * CI: Removing AKS engine related code (#2089) (cherry picked from commit b45c2c7) * feat: [dropgz] Dropgz for windows (#2075) * feat: [dropgz] Dropgz for windows * Removing the code for killing the process from dropgz for windows (cherry picked from commit 7a41178) * ci: Update dns tests for k8s conformance (#2104) Update dns tests for k8s v1.26 (cherry picked from commit bbf2fd4) * ci: adding cni package as a trigger (#2108) (cherry picked from commit e6a8ea6) * ci: add packages for submodule trigger (#2154) (cherry picked from commit 4aecfd6) * set mellanox reg key (#1768) (cherry picked from commit fa2de6d) --------- Co-authored-by: Evan Baker <rbtr@users.noreply.github.com> Co-authored-by: Camryn Lee <31013536+camrynl@users.noreply.github.com> Co-authored-by: Vipul Singh <vipul21sept@gmail.com> Co-authored-by: Rajvi <107083915+rajvinar@users.noreply.github.com>
Reason for Change:
Improves the test coverage for windows CNIv1 clusters.
Issue Fixed:
Requirements:
Notes: