Platform for building access networks and modular network services
This minor release for Magma contains fixes for some known issues affecting the stability and usability of the v1.3 and v1.3.1 releases of Durango. See the release page for more information.
Dec 04 00:24:42 mag1 systemd[1]: agw_installation.service: Failed with result 'timeout'.
deb http://packages.magma.etagecom.io stretch-stable main
AGWs should be configured to use deb http://packages.magma.etagecom.io stretch-1.3.2 main.
After this an apt update && apt upgrade magma
should install v1.3.2 on the gateway.Build new images https://magma.github.io/magma/docs/orc8r/deploy_build. Follow documentation to build and push packages to an image repository.
export
MAGMA_TAG=1.3.2-master
terraform init
terraform plan
terraform apply
Upgrade for AGW will be slightly different and will need to be done on the CLI.
‘lte/gateway/release/change_source.sh'
and run it as follows. The script will update your local apt repository to the latest one../change_source.sh 1.3.2
This should change the source.list.d entry for the AGW image.
1.3.2-1607014227-1462d960
into the tier that manages the AGWs. This change should be streamed to the device and magmad should start the upgrade process.apt upgrade magma
after running the change_source.sh
bash script noted above.Release 1.3.0 and 1.3.1 have been deprecated. New installs should use 1.3.2+
This hotfix release for Magma contains support for some new features and fixes for other known issues. See the release page for more information.
Magma now supports registering externally managed eNBs that aren’t configured by the AGW. This enables tracking of the state of the registered unmanaged eNBs as well as monitoring real time status and traffic usage.
Release 1.3.0 and 1.3.1 have been depreciated. New installs should use 1.3.2
v1.3.0 (Durango) is a major feature release. See https://magma.github.io/magma/docs/orc8r/upgrade_1_3 and https://magma.github.io/magma/docs/lte/agw_130_upgrade for upgrade instructions.
This release adds support for exporting Charging Data Records (CDRs) based on the IPFix (IP Flow Information Export) protocol. Magma also provides the following additional fields: imsi
, msisdn
, apn_name
, apn_mac
, pdp_start_time
, app_name
.
Magma now supports the Gx and Gy interfaces per 3GPP Diameter protocol for the FWA use-case. Notable new features are:
This release also includes several bug fixes to stabilize:
This feature enables the overriding of UE requested APN with a network specified APN via IMSI prefix based filtering. Up to 10 IMSI prefix filters and corresponding APNs to overwrite with can be defined.
This feature allows PLMN based filtering on subscribers that are allowed to authenticate with the configured HSS.
This new feature was added to support delivery of mobile-terminated SMS (i.e., receive-only SMS) to allow operators to deliver alerts to users. We have introduced a new mode of operation for the MME which is essentially “SMS-in-MME” mode in the 3GPP specification, with delivery coordinated by the Orchestrator. We mostly re-use infrastructure for the SGs interface used for CSFB support. Magma provides REST endpoints for provisioning SMS messages and checking their delivery status.
apn_resources
field cannot be updated while its connected_enodeb_serials
field is non-empty. To update the apn_resources
field in this situation, use the following workaround:
apn_resources
to an empty collection, with connected_enodeb_serials
as the desired valueapn_resources
to the desired value, and connected_enodeb_serials
as an empty collectionBoth of these known issues have fixes backported onto the release branch for a 1.3.1 bugfix release.
1.2.0 (Cotopaxi) is a major feature release. Manual upgrade steps will be required. See https://magma.github.io/magma/docs/orc8r/upgrade_1_2 and https://magma.github.io/magma/docs/lte/agw_120_upgrade for specific upgrade instructions.
With this version, we're rolling out a comprehensive redesign of the NMS. See the User Guide for an in-depth overview of the new system.
Note that quite a lot of new NMS functionality requires Elasticsearch to be deployed with Orchestrator. If you've turned off Elasticsearch in your Orchestrator deployment, this upgrade might be a good time to change that setting in your Terraform root module.
We've added a few new UE IP allocation strategies under "Bridged Mode" for the AGW. These include:
We are also rolling out multi APN-SGi support in this release as an experimental feature.
See the feature guide for bridged mode to use these new features.
Magma now supports configurable Quality of Service (QoS) profiles for traffic policies. You can now configure QoS parameters for each APN that a subscriber is allowed to connect with. Specific traffic policies with QoS parameters can also be assigned to subscribers on a per-APN basis to enforce finer-grained QoS restrictions.
In-depth feature documentation for QoS is still in the works, but you can refer to the original feature proposal for more details in the meantime.
1.1.0 (Bandai) is a feature release and will require some manual upgrade steps. See https://facebookincubator.github.io/magma/docs/orc8r/deploy_upgrade and https://facebookincubator.github.io/magma/docs/lte/agw_110_upgrade for upgrade instructions.
We've added multi-tenancy support to the LTE NMS. Tenants in the NMS are called "Organizations" and can own any subset of the networks that have been provisioned in Orchestrator. User accounts are now bound to specific organizations and are restricted from seeing any data from networks that their organization does not have access to.
Documentation on NMS organizations can be found at https://facebookincubator.github.io/magma/docs/nms/nms_organizations.
You can now create custom dashboards for timeseries data in the NMS using Grafana. Dashboards are scoped to organizations and any changes one user makes to a dashboard will be immediately visible to other users in the organization when saved.
Documentation on the new dashboards can be found at https://facebookincubator.github.io/magma/docs/nms/nms_grafana.
Starting with this release we are using Terraform for the entire Orchestrator deployment (infrastructure and Helm). This should be a much smoother deployment experience than the old procedure.
This release includes support for aggregating logs from Access Gateways into an Elasticsearch domain in AWS. By default, this option is turned off, but you can enable the Elasticsearch domain in the Terraform module if you'd like to use this. We are targeting the 1.2.0 release to integrate log aggregation with the NMS, so in this release the only way to view logs is directly in Kibana with no multi-tenant support.
Documentation on enabling this feature for individual access gateways will be added soon.
First upgrade your Orchestrator deployment following https://facebookincubator.github.io/magma/docs/orc8r/deploy_upgrade, then upgrade your gateways following https://facebookincubator.github.io/magma/docs/lte/agw_110_upgrade. The Orchestrator upgrade can be performed completely online with no traffic disruption to your AGWs, but the AGW upgrade will temporarily disable LTE service for that gateway.
v1.0.1 (Alto) is an important bugfix release for Magma. Access gateway builds on baremetal hardware are now reproducible - the install procedure in 1.0.0 would actually install versions of Magma from master instead of the 1.0.0 tag. Following the install directions in the 1.0.1 documentation will now deterministically install 1.0.1 on the access gateway.
See the "Release Details" section of the release notes for a complete list of all new commits that are in this release. The list of issues fixed in this release can be found under the 1.0.1 milestone on the issues page on Github.
You can follow the baremetal provisioning steps in the documentation to upgrade your AGW to 1.0.1, or from inside the AGW you can run:
sudo apt-get install -o Dpkg::Options::="--force-confnew" --assume-yes --force-yes magma=1.0.1-1580860031-33a94e98
If you ever used the NMS frontend to delete an access gateway from a network, you probably encountered #1071. To clear these "N/A" entries from your NMS, execute the following command after upgrading Orchestrator:
kubectl exec -n magma \
$(kubectl -n magma get pod -l app.kubernetes.io/component=controller -o jsonpath='{.items[0].metadata.name}') -- \
envdir /var/opt/magma/envdir /var/opt/magma/bin/m007_na_cleanup
This should log "SUCCESS". When you reload your NMS pages, you should find that the "N/A" gateway entries have disappeared.
Magma v1.0.0 (Alto) is the first stable release for fixed wireless and private LTE use cases. Refer to the documentation for an installation and deployment guide.
If you are on a 1.0.0 pre-release candidate, you will have to make the following changes to your Helm values.yml:
name: orc8r-bootstrap-legacy
to the proxy.service
sectionnode-selector
to nodeSelector
alertmanger
to alertmanager
Make the following change to your secrets:
envdir
should be STREAMER,SUBSCRIBERDB,METRICSD,CERTIFIER,BOOTSTRAPPER,METERINGD_RECORDS,ACCESSD,OBSIDIAN,DISPATCHER,DIRECTORYD
Rebuild your containers and upgrade your helm deployment afterwards to pick up the changes (don't forget to update the image tags in values.yml).
cellular
typefor
field in alert rule config (#629)originatingNetwork
label to orc8r alertsFirst release candidate of Magma's stable 1.0.0 release. The NMS has been refactored to use V1 API endpoints found at /apidocs/v1/. Our official recommendation is to only use these V1 API endpoints moving forward.
Known issues:
cellular
typefor
field in alert rule config (#629)originatingNetwork
label to orc8r alertsFirst alpha release of 1.0.0. From 1.0.0 forwards, our commitment will be that the API endpoints under /apidocs/v1/
will always be backwards-compatible within a major version (i.e. 1.x.y).
All major private LTE/fixed wireless API endpoints are implemented as far as 1.0.0-alpha1. We recommend only using the endpoints found at /apidocs/v1/
, not the ones at /apidocs
from here. We will support both APIs until the magmalte
NMS migrates to the V1 endpoints, at which point we will remove support for the legacy endpoints entirely.