Signed-off-by: Eli Sutton [email protected]

Summary

Make state service sync interval configurable from orchestrator Issue #8482

Test Plan

Unit Tests Local testing

Additional Information

Test case :

1. verified basic flows using UERANSIM a) Initial Registration b) PDUSession c) Deregistration
2. Run all mme unit test cases

Signed-off-by: ganeshg87 [email protected]

Summary

• 5G Core interactions - Access and Mobility Function(AMF) included(amf)
• NGAP changes for 5G gNB interactions
• n11 interface between AMF - SMF
• 5G Authentication feature.

Test Plan

Additional Information

• [ ] This change is backwards-breaking

Your Environment

• Version: v1.3 stable (patched MME and patched uplink_bridge.py)
• Affected Component: Orchestrator & Access Gateway
• Affected Subcomponent: SMSd and MME
• Deployment Environment: e.g. AWS (orc8r) , bare-metal (AGW), iPhone 11pro

Describe the Issue

Trying to test SMS functionality via SMSd. I first used API to try to send SMS to IMSI, but it returns 0 attempts and status waiting. (see screenshot)

I then tried to use the sms_cli to send the SMS, but that appears to crash the mme.

To Reproduce

1. Attach iPhone 11pro to Network
2. Via SSH in AGW, issue sms_cli du mesage to UE e.g.sms_cli.py DU 315010999900110 1fc13a00
3. Observe journal where MME sig terms and restarts

Expected behavior Expected behavior is SMS is delivered to phone or fails without taking down mme.

Screenshots Screenshot of API SMS none-delivery

Additional context ZIp of: console log of command pcap of s1 mme.log journal of sequence of UE attach, issue sms_cli.py command, mme sig term 10:29 Testing.zip

Test Network drawing (orc8r in AWS)

Summary

+--------------------------------------+ Note: Merge together with #12422.
+--------------------------------------+

This PR upgrades Magma to Go 1.18. While there are various good reasons to upgrade Go to a recent version, the initial motivation for us was the implementation of the distributed tracing proposal (#10492). The needed Open Telemetry modules rely on Go >= 1.16.

Closes https://github.com/magma/magma/issues/4055, closes https://github.com/magma/magma/issues/12559.

We recommend to review commit by commit. (Many changes are in generated code, which we put into the ./build.py --generate commit).

Most prominent changes

go-swagger

With Go various other modules have to be upgraded as well. The most influential change is the version bump in go-swagger from v0.21.0 to v0.29.0 (currently latest). It was necessary to get swagger generation going.

The upgrade in go-swagger causes changes in generated files and requires minor changes in code and tests at various places to maintain previous functionality. They are bundled in the following commits:

• snapshot tests: RewriteTest and GenerateTest: substitute expected...
• fix build and test errors due to swagger generation...

go install and go get

Behaviour of go install and go get changes in Go >= 1.16 (see https://go.dev/doc/go1.16 for go install and https://go.dev/doc/go1.18 for go get).

If used within a module, a version suffix has to be used to ignore module requirements (e.g. go install [email protected]). Without version suffix an entry in go.mod file is required. This causes commands like make tools to fail with Go >=1.16. We thus updated existing tools.go files and introduced new tools.go files in commit Update existing tools.go files...

Commit Replace go get with go install... is due to changes of go get behaviour.

Dockerfiles

Dockerfiles are updated to use Go 1.18 in commits Update Dockerfiles to Go 1.18 and update orc8r_deployer dockerfile....

VM and DevContainer can be upgraded independently once this PR is merged (see #12422 and #12424).

CI

Commit Update Go to 1.18 in CI switches to Go 1.18 in various workflows. We also filed two separate PRs with CI Updates in #12352 and #12305 as prerequisite for this PR.

mockery and go-acc

An upgrade of mockery to v2 and go-acc to v0.2.8 is also necessary and done in commits upgrade mockery to v2... and upgrade go-acc to get it running.... Since we changed some mocked functions to use pointers commit generate new mockery files is needed.

go-openapi in cwf_operator module

To fix an error similar to this one https://github.com/kubernetes-sigs/kustomize/tree/master/cmd/depprobcheck certain versions of go-openapi modules were pinned in commit Replace versions of go-openapi/spec… to resolve incomaptible versions.

Test Plan

• [x] CI (Also ran various checks locally, but they are all included in the CI.)
• [x] local tests with running orc8r, AGW and NMS: tested various workflows like
  • end2end workflow of updating config yaml file in NMS
  • setting configs with swagger
  • propagation of mconfig to AGW
• [x] Precommit checks in feg/gateway/docker and orc8r/cloud/docker, ./build.py -c
• [x] cwag precommit test, vagrant up cwag && vagrant ssh cwag && cd magma/cwf/gateway && make precommit (relies on first doing vagrant up cwag --provision with https://github.com/magma/magma/pull/12422)
• [x] cwag integ test: https://docs.magmacore.org/docs/cwf/dev_testing#run-integration-tests
• [x] Federated integ test: https://docs.magmacore.org/docs/next/feg/s1ap_federated_tests#s1ap-federated-integration-tests
• [x] Build and deploy a K8s Orc8r (can be locally using Minikube)

Open point for follow-up PR

• Run make gen in orc8r/cloud/api/v1/. This generates changes in roughly 1000 files, might be easier to review this one if done in a seprate PR.

Additional Information

• Done in pairing with @sebathomas, @voisey, @ajahl, @Neudrino.

• Once this PR is merged you will have to run vagrant up magma --provision to install Go 1.18 on your VM.

• [ ] This change is backwards-breaking

Brief Context & scope

We would like to have the Domain proxy integration tests with orc8r Check from dp-workflow required in CI. Changes in orc8r possibly can have an impact on the Domain Proxy component, we would like to avoid merging changes which break compatibility between Orchestrator and Domain Proxy. The Check was improved and recently is robust and we didn't see flakiness in this job

Link to job definition: https://github.com/magma/magma/blob/7fe7fde1607dd015c4210b2d70de0f8efacd78d8/.github/workflows/dp-workflow.yml#L370

Goals

Prevent any PRs that break this test from being merged into master.

Proposal

• Make an announcement in #dev that this test will now be mandatory.
• Wait 3 days for any objections
• Turn on the required check!
• Update documentation on GitHub wiki - move check from "non-blocking" to "blocking" table

Your Environment

• Version: 1.5
• Affected Component: NMS
• Affected Subcomponent: AGW Health in NMS
• Deployment Environment: AWS (orc8r)

Describe the Issue

Customers see AGW health shown as BAD in NMS but becomes good after refreshing the page Internal Task: T93533003

Sync RPC has historically caused issues with federated deployments. Here we intend to root cause and fix this issue.

The issue appears most frequently when load testing a federated deployment. Above 1k attaches (producing s6a and session proxy RPC calls), the underlying h2 connection gets closed. This causes all calls through feg relay to fail until magmad on the feg can initiate a new sync RPC connection. It is worth noting that orc8r's shift from nghttpx to nginx as made the issue worse.

Signed-off-by: GANESH IRRINKI [email protected]

Summary

• 3GPP SBI Code generated based on https://github.com/deepmap/oapi-codegen. SBI openapis are defined in OpenAPI 3.0 specification and go-swagger will not work.
• Though only N7 inteface openapi is required, it has depencies on other 3GPP openapi files. Hence other files are also included.
• The OpenAPI specifications are taken from 3GPP specification v15.8. The following modifications are made to the specifications
  • Resolved cycling dependencies by moving the definitions to common openapi file
  • Changed few names that were causing compilation error on the generated code
• Added go:generate for regenerating the go code, when the openapi yaml files are changed.
• Each generated go file is maintained in its own package, because of the way the code is generated and to avoid conflicting definitions.

Files obtained from 3GPP are in the directory - feg/gateway/sbi/defs/TS*.yaml

Files generated - feg/gateway/sbi/specs/TS*/TS*.gen.go

Files modified or newly added feg/gateway/sbi/sbi_test.go feg/gateway/sbi/sbi.go feg/gateway/docker/go/Dockerfile feg/gateway/go.mod feg/gateway/go.sum feg/gateway/Makefile

Test Plan

Unit test code tests client request to a mock server.

Additional Information

feat(agw): Added stateless support for ueip_imsi map

Summary

1. Added stateless support for ueip_imsi map
2. Modified state_cli.py script to parse mme_ueip_imsi_map
3. Modified config_stateless_agw.py script to clear contents of mme_ueip_imsi_map
4. Added new test case, test_paging_after_mme_restart.py to validate the resumption of mme_ueip_imsi_map after mme has restarted and added to sanity test suite
5. Modified s1ap_utils.py to get the remaining entries after test case is completed.

Test Plan

Executed s1ap sanity test suite

Your Environment

• Version: 1.3.2 (with patch to lte/gateway/c/oai/tasks/nas/emm/sap/emm_send.c based on e5638c7)
• Affected Component: Access Gateway
• Affected Subcomponent: MME/Mobilityd/pipelined
• Deployment Environment: bare-metal (AGW)

Describe the Issue

Got a customer call about outage of MME causing all UEs to lose connection for short period. (i.e. mme restart)

Looking at logs I can't see what went wrong. At 15:20:13 we start seeing error message from mobilityd about IP address allocation issue. Odd because IP are statically assigned for all SIM cards. Note all the notices about subscriberDB being locked in proximity to that time in syslog.

Network is per this diagram (bridge mode, static IPs everywhere WAN, UEs, eth1, eNB etc).

There is a core-* file produced. But our attempts to open with GDB it complains about the binary not matching even when we use the matching mme binary (both attached).

removed the logs and shared it privately

Hi Arun,

Please arrange the automation TC for Multi session to cover required positive and negative scenarios.

I created TC in the requested format. Please find the attached document for your perusal and let me know for any additional inputs.

https://wavelabstechnologies.sharepoint.com/:x:/r/sites/Wavelabs-Connectivity/_layouts/15/Doc.aspx?sourcedoc=%7B119DE675-716C-4565-8216-8DEBB2489931%7D&file=Feature_TestCase_Desing_Document.xlsx&action=default&mobileredirect=true&wdOrigin=TEAMS-ELECTRON.teams.files&wdExp=TEAMS-CONTROL&wdhostclicktime=1637324150718

Thanks and Regards, Neelima.

Summary

Docker is deprecating docker compose V1. Additionally, the new features of docker compose V2 may avoid a lot of manual workarounds and mitigations for unavailable functionality in V1. Since Docker changed the naming scheme of their images we run the upgrade version in a compatibility mode, denoted with several --compatibility flags. This allows us to keep the current naming scheme and docker-compose files.

Linked to https://github.com/magma/magma/issues/13938.

Test Plan

• [x] Build Orc8r, SwaggerAPI, NMS along the Quick Start Guide
• [x] Build containerized AGW along the (adapted) README.md
• [x] LTE integration tests (incl. traffic server tests) w/ containerized AGW
  • [x] precommit
  • [x] extended with failing test_attach_detach_rar_tcp_he.py and test_attach_detach_with_he_policy.py -> Same as in this master run.
  • [x] extended long
• [x] CWAG build and integration test suite
  • Run 1 with failing TestGyCreditExhaustionWithoutCCRU
  • Run 2 with failing TestAuthenticateUplinkTraffic and TestGxReAuthWithMidSessionPolicyRemoval
• [x] FEG integration test suite
• [x] Dev Container
• Most GitHub Actions / CI only to be seen in CI / Artifactory

Additional Information

:warning: To start with a clean slate, this PR is based on the revert PR #14795.

Signed-off-by: Marco Pfirrmann [email protected]

Summary

Instead of container digests, we select containers with their name (which consists of the corresponding commit hash / latest tag) to run containerized integ tests with. While this does not matter for CI, it improves the interface for manually executing these tests since we now provide a default value for containers. Closes 14798.

Test Plan

• Containerized integ test with default (=latest) parameters

Additional Information

• [ ] This change is backwards-breaking

Signed-off-by: Nils Semmelrock [email protected]

Summary

This change tries to tackle multiple issues. The main idea is to not fetch the current datapath during static initialization of the s1ap_utils, but in the constructor when an object is created.

Tackling #14068

• Moving get_dapatath to the constructor of the s1ap_utils makes it possible to wait (before the s1ap_utils are initialized) for pipelined to actually initialize a datapath
• this should mitigate issues described in #14068
• for this the initialization of the s1ap_utils needs to be moved after the magmad_util is initialized (holds the functionality to wait for the start of pipelined)

Working towards enabling again that was disable in #14682

• With the s1ap_utils initalization moved after initializing magmad_util we can check if nat is not enabled - and enabling it in this case (this can happen if a test that run with disabled nat failed with an error that prevents the re-enabling of nat in the teardown).
• If a test with disabled nat tries to call pipelined, then we see 6h workflow timeouts in ci.

Tackling #14699

• This is basically the main issue that tracks the two issues above.

Test Plan

executed precommit and extended tests locally

Additional Information

• [ ] This change is backwards-breaking

Signed-off-by: Nils Semmelrock [email protected]

Summary

This change adds integ test results for the lte containerized tests to the github workflow overview page. This makes it more easy to quickly identify failing tests.

Test Plan

CI

Additional Information

• [ ] This change is backwards-breaking

Signed-off-by: Nils Semmelrock [email protected]

Summary

Small change to remove a duplicated step in the bazel debian integ tests - this had, as it seems, no negative effect, i.e., this is only clean-up.

Test Plan

CI

Additional Information

• [ ] This change is backwards-breaking