Welcome to Knowledge Base!

KB at your finger tips

Book a Meeting to Avail the Services of GitLab overtime

This is one stop global knowledge base where you can learn about all the products, solutions and support features.

Categories
All

GitLab

Use TLS between components of the GitLab chart | GitLab







  • Preparation


    • Generating certificates for internal use

      • Required certificate CN and SANs
  • Configuration
  • Result
  • Troubleshooting

Use TLS between components of the GitLab chart

The GitLab charts can use transport-layer security (TLS) between the various
components. This requires you to provide certificates for the services
you want to enable, and configure those services to make use of those
certificates and the certificate authority (CA) that signed them.

Preparation

Each chart has documentation regarding enabling TLS for that service, and the various
settings required to ensure that appropriate configuration.

Generating certificates for internal use


note
GitLab does not purport to provide high-grade PKI infrastructure, or certificate
authorities.

For the purposes of this documentation, we provide a Proof of Concept script
below, which makes use of Cloudflare’s CFSSL
to produce a self-signed Certificate Authority, and a wildcard certificate that can be
used for all services.

This script will:


  • Generate a CA key pair.
  • Sign a certificate meant to service all GitLab component service endpoints.
  • Create two Kubernetes Secret objects:

    • A secret of type kuberetes.io/tls which has the server certificate and key pair.
    • A secret of type Opaque which only contains the public certificate of the CA as ca.crt
      as need by NGINX Ingress.

Prerequisites:


  • Bash, or compatible shell.

  • cfssl is available to your shell, and within PATH .

  • kubectl is available, and configured to point to your Kubernetes cluster
    where GitLab will later be installed.

    • Be sure to have created the namespace you wish to have these certificates
      installed into before operating the script.

You may copy the content of this script to your computer, and make the resulting
file executable. We suggest poc-gitlab-internal-tls.sh .

#!/bin/bash
set -e
#############
## make and change into a working directory
pushd $(mktemp -d)

#############
## setup environment
NAMESPACE=${NAMESPACE:-default}
RELEASE=${RELEASE:-gitlab}
## stop if variable is unset beyond this point
set -u
## known expected patterns for SAN
CERT_SANS="*.${NAMESPACE}.svc,${RELEASE}-metrics.${NAMESPACE}.svc,*.${RELEASE}-gitaly.${NAMESPACE}.svc"

#############
## generate default CA config
cfssl print-defaults config > ca-config.json
## generate a CA
echo '{"CN":"'${RELEASE}.${NAMESPACE}.internal.ca'","key":{"algo":"ecdsa","size":256}}' | \
cfssl gencert -initca - | \
cfssljson -bare ca -
## generate certificate
echo '{"CN":"'${RELEASE}.${NAMESPACE}.internal'","key":{"algo":"ecdsa","size":256}}' | \
cfssl gencert -config=ca-config.json -ca=ca.pem -ca-key=ca-key.pem -profile www -hostname="${CERT_SANS}" - |\
cfssljson -bare ${RELEASE}-services

#############
## load certificates into K8s
kubectl -n ${NAMESPACE} create secret tls ${RELEASE}-internal-tls \
--cert=${RELEASE}-services.pem \
--key=${RELEASE}-services-key.pem
kubectl -n ${NAMESPACE} create secret generic ${RELEASE}-internal-tls-ca \
--from-file=ca.crt=ca.pem

note
This script does not preserve the CA’s private key. It is a Proof-of-Concept
helper, and is not intended for production use .

The script expects two environment variables to be set:



  1. NAMESPACE : The Kubernetes Namespace you will later install GitLab to.
    This defaults to default , as with kubectl .

  2. RELEASE : The Helm Release name you will later use to install GitLab.
    This defaults to gitlab .

To operate this script, you may export the two variables, or prepend the
script name with their values.

export NAMESPACE=testing
export RELEASE=gitlab

./poc-gitlab-internal-tls.sh

After the script has run, you will find the two secrets created, and the
temporary working directory contains all certificates and their keys.

$ pwd
/tmp/tmp.swyMgf9mDs
$ kubectl -n ${NAMESPACE} get secret | grep internal-tls
testing-internal-tls kubernetes.io/tls 2 11s
testing-internal-tls-ca Opaque 1 10s
$ ls -1
ca-config.json
ca.csr
ca-key.pem
ca.pem
testing-services.csr
testing-services-key.pem
testing-services.pem

Required certificate CN and SANs

The various GitLab components speak to each other over their Service’s DNS names.
The Ingress objects generated by the GitLab chart must provide NGINX the
name to verify, when tls.verify: true (which is the default). As a result
of this, each GitLab component should receive a certificate with a SAN including
either their Service’s name, or a wildcard acceptable to the Kubernetes Service
DNS entry.


  • service-name.namespace.svc
  • *.namespace.svc

Failure to ensure these SANs within certificates will result in a non-functional
instance, and logs that can be quite cryptic, refering to “connection failure”
or “SSL verification failed”.

You can make use of helm template to retrieve a full list of all
Service object names, if needed. If your GitLab has been deployed without TLS,
you can query Kubernetes for those names:

kubectl -n ${NAMESPACE} get service -lrelease=${RELEASE}

Configuration

Example configurations can be found in examples/internal-tls.

For the purposes of this documentation, we have provided shared-cert-values.yaml
which configures the GitLab components to consume the certificates generated with
the script above, in generating certificates for internal use.

Key items to configure:


  1. Global Custom Certificate Authorities.
  2. Per-component TLS for the service listeners.
    (See each chart’s documentation, under charts/)

This process is greatly simplified by making use of YAML’s native anchor
functionality. A truncated snippet of shared-cert-values.yaml shows this:

.internal-ca: &internal-ca gitlab-internal-tls-ca
.internal-tls: &internal-tls gitlab-internal-tls

global:
certificates:
customCAs:
- secret: *internal-ca
workhorse:
tls:
enabled: true
gitlab:
webservice:
tls:
secretName: *internal-tls
workhorse:
tls:
verify: true # default
secretName: *internal-tls
caSecretName: *internal-ca

Result

When all components have been configured to provide TLS on their service
listeners, all communication between GitLab components will traverse the
network with TLS security, including connections from NGINX Ingress to
each GitLab component.

NGINX Ingress will terminate any inbound TLS, determine the appropriate
services to pass the traffic to, and then form a new TLS connection to
the GitLab component. When configured as shown here, it will also verify
the certificates served by the GitLab components against the CA.

This can be verified by connecting to the Toolbox pod, and querying the
various compontent Services. One such example, connecting to the Webservice
Pod’s primary service port that NGINX Ingress uses:

$ kubectl -n ${NAMESPACE} get pod -lapp=toolbox,release=${RELEASE}
NAME READY STATUS RESTARTS AGE
gitlab-toolbox-5c447bfdb4-pfmpc 1/1 Running 0 65m
$ kubectl exec -ti gitlab-toolbox-5c447bfdb4-pfmpc -c toolbox -- \
curl -Iv "https://gitlab-webservice-default.testing.svc:8181"

The output should be similar to following example:

*   Trying 10.60.0.237:8181...
* Connected to gitlab-webservice-default.testing.svc (10.60.0.237) port 8181 (#0)
* ALPN, offering h2
* ALPN, offering http/1.1
* successfully set certificate verify locations:
* CAfile: /etc/ssl/certs/ca-certificates.crt
* CApath: /etc/ssl/certs
* TLSv1.3 (OUT), TLS handshake, Client hello (1):
* TLSv1.3 (IN), TLS handshake, Server hello (2):
* TLSv1.3 (IN), TLS handshake, Encrypted Extensions (8):
* TLSv1.3 (IN), TLS handshake, Certificate (11):
* TLSv1.3 (IN), TLS handshake, CERT verify (15):
* TLSv1.3 (IN), TLS handshake, Finished (20):
* TLSv1.3 (OUT), TLS change cipher, Change cipher spec (1):
* TLSv1.3 (OUT), TLS handshake, Finished (20):
* SSL connection using TLSv1.3 / TLS_AES_128_GCM_SHA256
* ALPN, server did not agree to a protocol
* Server certificate:
* subject: CN=gitlab.testing.internal
* start date: Jul 18 19:15:00 2022 GMT
* expire date: Jul 18 19:15:00 2023 GMT
* subjectAltName: host "gitlab-webservice-default.testing.svc" matched cert's "*.testing.svc"
* issuer: CN=gitlab.testing.internal.ca
* SSL certificate verify ok.
> HEAD / HTTP/1.1
> Host: gitlab-webservice-default.testing.svc:8181

Troubleshooting

If your GitLab instance appears unreachable from the browser, by rendering an
HTTP 503 error, NGINX Ingress is likely having a problem verifying the
certificates of the GitLab components.

You may work around this by temporarily setting
gitlab.webservice.workhorse.tls.verify to false .

The NGINX Ingress controller can be connected to, and will evidence a message
in nginx.conf , regarding problems verifying the certificate(s).

Example content, where the Secret is not reachable:

# Location denied. Reason: "error obtaining certificate: local SSL certificate
testing/gitlab-internal-tls-ca was not found"
return 503;

Common problems that cause this:


  • CA certificate is not in a key named ca.crt within the Secret.
  • The Secret was not properly supplied, or may not exist within the Namespace.

Stay Ahead in Today’s Competitive Market!
Unlock your company’s full potential with a Virtual Delivery Center (VDC). Gain specialized expertise, drive seamless operations, and scale effortlessly for long-term success.

Book a Meeting to Avail the Services of GitLabovertime

Configure the GitLab chart with persistent volumes | GitLab






  • Locate the GitLab Volumes
  • Before making storage changes

  • Making storage changes


    • Changes to an existing Volume

      • Update the volume to bind to the claim
    • Switching to a different Volume
  • Make changes to the PersistentVolumeClaim
  • Apply the changes to the GitLab chart

Configure the GitLab chart with persistent volumes

Some of the included services require persistent storage, configured through
Persistent Volumes that specify which disks your cluster has access to.
Documentation on the storage configuration necessary to install this chart can be found in our
Storage Guide.

Storage changes after installation need to be manually handled by your cluster
administrators. Automated management of these volumes after installation is not
handled by the GitLab chart.

Examples of changes not automatically managed after initial installation
include:


  • Mounting different volumes to the Pods
  • Changing the effective accessModes or Storage Class
  • Expanding the storage size of your volume* 1

1 In Kubernetes 1.11, expanding the storage size of your volume is supported
if you have allowVolumeExpansion configured to true in your Storage Class.

Automating theses changes is complicated due to:


  1. Kubernetes does not allow changes to most fields in an existing PersistentVolumeClaim
  2. Unless manually configured, the PVC is the only reference to dynamically provisioned PersistentVolumes

  3. Delete is the default reclaimPolicy for dynamically provisioned PersistentVolumes

This means in order to make changes, we need to delete the PersistentVolumeClaim
and create a new one with our changes. But due to the default reclaimPolicy,
deleting the PersistentVolumeClaim may delete the PersistentVolumes
and underlying disk. And unless configured with appropriate volumeNames and/or
labelSelectors, the chart doesn’t know the volume to attach to.

We will continue to look into making this process easier, but for now a manual
process needs to be followed to make changes to your storage.

Locate the GitLab Volumes

Find the volumes/claims that are being used:

kubectl --namespace <namespace> get PersistentVolumeClaims -l release=<chart release name> -ojsonpath='{range .items[*]}{.spec.volumeName}{"\t"}{.metadata.labels.app}{"\n"}{end}'


  • <namespace> should be replaced with the namespace where you installed the GitLab chart.

  • <chart release name> should be replaced with the name you used to install the GitLab chart.

The command prints a list of the volume names, followed by the name of the
service they are for.

For example:

$ kubectl --namespace helm-charts-win get PersistentVolumeClaims -l release=review-update-app-h8qogp -ojsonpath='{range .items[*]}{.spec.volumeName}{"\t"}{.metadata.labels.app}{"\n"}{end}'
pvc-6247502b-8c2d-11e8-8267-42010a9a0113 gitaly
pvc-61bbc05e-8c2d-11e8-8267-42010a9a0113 minio
pvc-61bc6069-8c2d-11e8-8267-42010a9a0113 postgresql
pvc-61bcd6d2-8c2d-11e8-8267-42010a9a0113 prometheus
pvc-61bdf136-8c2d-11e8-8267-42010a9a0113 redis

Before making storage changes

The person making the changes needs to have administrator access to the cluster, and appropriate access to the storage
solutions being used. Often the changes will first need to be applied in the storage solution, then the results need to
be updated in Kubernetes.

Before making changes, you should ensure your PersistentVolumes are using
the Retain reclaimPolicy so they don’t get removed while you are
making changes.

First, find the volumes/claims that are being used.

Next, edit each volume and change the value of persistentVolumeReclaimPolicy
under the spec field, to be Retain rather than Delete

For example:

kubectl --namespace helm-charts-win edit PersistentVolume pvc-6247502b-8c2d-11e8-8267-42010a9a0113

Editing Output:

# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file will be
# reopened with the relevant failures.
#
apiVersion: v1
kind: PersistentVolume
metadata:
annotations:
kubernetes.io/createdby: gce-pd-dynamic-provisioner
pv.kubernetes.io/bound-by-controller: "yes"
pv.kubernetes.io/provisioned-by: kubernetes.io/gce-pd
creationTimestamp: 2018-07-20T14:58:43Z
labels:
failure-domain.beta.kubernetes.io/region: europe-west2
failure-domain.beta.kubernetes.io/zone: europe-west2-b
name: pvc-6247502b-8c2d-11e8-8267-42010a9a0113
resourceVersion: "48362431"
selfLink: /api/v1/persistentvolumes/pvc-6247502b-8c2d-11e8-8267-42010a9a0113
uid: 650bd649-8c2d-11e8-8267-42010a9a0113
spec:
accessModes:
- ReadWriteOnce
capacity:
storage: 50Gi
claimRef:
apiVersion: v1
kind: PersistentVolumeClaim
name: repo-data-review-update-app-h8qogp-gitaly-0
namespace: helm-charts-win
resourceVersion: "48362307"
uid: 6247502b-8c2d-11e8-8267-42010a9a0113
gcePersistentDisk:
fsType: ext4
pdName: gke-cloud-native-81a17-pvc-6247502b-8c2d-11e8-8267-42010a9a0113
# Changed the following line
persistentVolumeReclaimPolicy: Retain
storageClassName: standard
status:
phase: Bound

Making storage changes

First, make the desired changes to the disk outside of the cluster. (Resize the
disk in GKE, or create a new disk from a snapshot or clone, etc).

How you do this, and whether or not it can be done live, without downtime, is
dependent on the storage solutions you are using, and can’t be covered by this
document.

Next, evaluate whether you need these changes to be reflected in the Kubernetes
objects. For example: with expanding the disk storage size, the storage size
settings in the PersistentVolumeClaim will only be used when a new volume
resource is requested. So you would only need to increase the values in the
PersistentVolumeClaim if you intend to scale up more disks (for use in
additional Gitaly pods).

If you do need to have the changes reflected in Kubernetes, be sure that you’ve
updated your reclaim policy on the volumes as described in the Before making storage changes
section.

The paths we have documented for storage changes are:


  • Changes to an existing Volume
  • Switching to a different Volume

Changes to an existing Volume

First locate the volume name you are changing.

Use kubectl edit to make the desired configuration changes to the volume. (These changes
should only be updates to reflect the real state of the attached disk)

For example:

kubectl --namespace helm-charts-win edit PersistentVolume pvc-6247502b-8c2d-11e8-8267-42010a9a0113

Editing Output:

# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file will be
# reopened with the relevant failures.
#
apiVersion: v1
kind: PersistentVolume
metadata:
annotations:
kubernetes.io/createdby: gce-pd-dynamic-provisioner
pv.kubernetes.io/bound-by-controller: "yes"
pv.kubernetes.io/provisioned-by: kubernetes.io/gce-pd
creationTimestamp: 2018-07-20T14:58:43Z
labels:
failure-domain.beta.kubernetes.io/region: europe-west2
failure-domain.beta.kubernetes.io/zone: europe-west2-b
name: pvc-6247502b-8c2d-11e8-8267-42010a9a0113
resourceVersion: "48362431"
selfLink: /api/v1/persistentvolumes/pvc-6247502b-8c2d-11e8-8267-42010a9a0113
uid: 650bd649-8c2d-11e8-8267-42010a9a0113
spec:
accessModes:
- ReadWriteOnce
capacity:
# Updated the storage size
storage: 100Gi
claimRef:
apiVersion: v1
kind: PersistentVolumeClaim
name: repo-data-review-update-app-h8qogp-gitaly-0
namespace: helm-charts-win
resourceVersion: "48362307"
uid: 6247502b-8c2d-11e8-8267-42010a9a0113
gcePersistentDisk:
fsType: ext4
pdName: gke-cloud-native-81a17-pvc-6247502b-8c2d-11e8-8267-42010a9a0113
persistentVolumeReclaimPolicy: Retain
storageClassName: standard
status:
phase: Bound

Now that the changes have been reflected in the volume, we need to update
the claim.

Follow the instructions in the Make changes to the PersistentVolumeClaim section.

Update the volume to bind to the claim

In a separate terminal, start watching to see when the claim has its status change to bound,
and then move onto the next step to make the volume available for use in the new claim.

kubectl --namespace <namespace> get --watch PersistentVolumeClaim <claim name>

Edit the volume to make it available to the new claim. Remove the .spec.claimRef section.

kubectl --namespace <namespace> edit PersistentVolume <volume name>

Editing Output:

# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file will be
# reopened with the relevant failures.
#
apiVersion: v1
kind: PersistentVolume
metadata:
annotations:
kubernetes.io/createdby: gce-pd-dynamic-provisioner
pv.kubernetes.io/bound-by-controller: "yes"
pv.kubernetes.io/provisioned-by: kubernetes.io/gce-pd
creationTimestamp: 2018-07-20T14:58:43Z
labels:
failure-domain.beta.kubernetes.io/region: europe-west2
failure-domain.beta.kubernetes.io/zone: europe-west2-b
name: pvc-6247502b-8c2d-11e8-8267-42010a9a0113
resourceVersion: "48362431"
selfLink: /api/v1/persistentvolumes/pvc-6247502b-8c2d-11e8-8267-42010a9a0113
uid: 650bd649-8c2d-11e8-8267-42010a9a0113
spec:
accessModes:
- ReadWriteOnce
capacity:
storage: 100Gi
gcePersistentDisk:
fsType: ext4
pdName: gke-cloud-native-81a17-pvc-6247502b-8c2d-11e8-8267-42010a9a0113
persistentVolumeReclaimPolicy: Retain
storageClassName: standard
status:
phase: Released

Shortly after making the change to the Volume, the terminal watching the claim status should show Bound .

Finally, apply the changes to the GitLab chart

Switching to a different Volume

If you want to switch to using a new volume, using a disk that has a copy of the
appropriate data from the old volume, then first you need to create the new
Persistent Volume in Kubernetes.

In order to create a Persistent Volume for your disk, you will need to
locate the driver specific documentation
for your storage type.

There are a couple of things to keep in mind when following the driver documentation:


  • You need to use the driver to create a Persistent Volume, not a Pod object with a volume as shown in a lot of the documentation.
  • You do not want to create a PersistentVolumeClaim for the volume, we will be editing the existing claim instead.

The driver documentation often includes examples for using the driver in a Pod, for example:

apiVersion: v1
kind: Pod
metadata:
name: test-pd
spec:
containers:
- image: k8s.gcr.io/test-webserver
name: test-container
volumeMounts:
- mountPath: /test-pd
name: test-volume
volumes:
- name: test-volume
# This GCE PD must already exist.
gcePersistentDisk:
pdName: my-data-disk
fsType: ext4

What you actually want, is to create a Persistent Volume, like so:

apiVersion: v1
kind: PersistentVolume
metadata:
name: test-volume
spec:
capacity:
storage: 400Gi
accessModes:
- ReadWriteOnce
gcePersistentDisk:
pdName: my-data-disk
fsType: ext4

You normally create a local yaml file with the PersistentVolume information,
then issue a create command to Kubernetes to create the object using the file.

kubectl --namespace <your namespace> create -f <local-pv-file>.yaml

Once your volume is created, you can move on to Making changes to the PersistentVolumeClaim

Make changes to the PersistentVolumeClaim

Find the PersistentVolumeClaim you want to change.

kubectl --namespace <namespace> get PersistentVolumeClaims -l release=<chart release name> -ojsonpath='{range .items[*]}{.metadata.name}{"\t"}{.metadata.labels.app}{"\n"}{end}'


  • <namespace> should be replaced with the namespace where you installed the GitLab chart.

  • <chart release name> should be replaced with the name you used to install the GitLab chart.

The command will print a list of the PersistentVolumeClaim names, followed by the name of the
service they are for.

Then save a copy of the claim to your local filesystem:

kubectl --namespace <namespace> get PersistentVolumeClaim <claim name> -o yaml > <claim name>.bak.yaml

Example Output:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
annotations:
pv.kubernetes.io/bind-completed: "yes"
pv.kubernetes.io/bound-by-controller: "yes"
volume.beta.kubernetes.io/storage-provisioner: kubernetes.io/gce-pd
creationTimestamp: 2018-07-20T14:58:38Z
labels:
app: gitaly
release: review-update-app-h8qogp
name: repo-data-review-update-app-h8qogp-gitaly-0
namespace: helm-charts-win
resourceVersion: "48362433"
selfLink: /api/v1/namespaces/helm-charts-win/persistentvolumeclaims/repo-data-review-update-app-h8qogp-gitaly-0
uid: 6247502b-8c2d-11e8-8267-42010a9a0113
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 50Gi
storageClassName: standard
volumeName: pvc-6247502b-8c2d-11e8-8267-42010a9a0113
status:
accessModes:
- ReadWriteOnce
capacity:
storage: 50Gi
phase: Bound

Create a new YAML file for a new PVC object. Have it use the same metadata.name , metadata.labels , metadata,namespace , and spec fields. (With your updates applied). And drop the other settings:

Example:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
labels:
app: gitaly
release: review-update-app-h8qogp
name: repo-data-review-update-app-h8qogp-gitaly-0
namespace: helm-charts-win
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
# This is our updated field
storage: 100Gi
storageClassName: standard
volumeName: pvc-6247502b-8c2d-11e8-8267-42010a9a0113

Now delete the old claim:

kubectl --namespace <namespace> delete PersistentVolumeClaim <claim name>

Create the new claim:

kubectl --namespace <namespace> create PersistentVolumeClaim -f <new claim yaml file>

If you are binding to the same PersistentVolume that was previous bound to
the claim, then proceed to update the volume to bind to the claim

Otherwise, if you have bound the claim to a new volume, move onto apply the changes to the GitLab chart

Apply the changes to the GitLab chart

After making changes to the PersistentVolumes and PersistentVolumeClaims,
you will also want to issue a Helm update with the changes applied to the chart
settings as well.

See the installation storage guide
for the options.


Note : If you made changes to the Gitaly volume claim, you will need to delete the
Gitaly StatefulSet before you will be able to issue a Helm update. This is
because the StatefulSet’s Volume Template is immutable, and cannot be changed.

You can delete the StatefulSet without deleting the Gitaly Pods:
kubectl --namespace <namespace> delete --cascade=false StatefulSet <release-name>-gitaly
The Helm update command will recreate the StatefulSet, which will adopt and
update the Gitaly pods.

Update the chart, and include the updated configuration:

Example:

helm upgrade --install review-update-app-h8qogp gitlab/gitlab \
--set gitlab.gitaly.persistence.size=100Gi \
<your other config settings>
Read article

Configure the GitLab chart with UBI-based images | GitLab






  • Sample values
  • Known Limitations

Configure the GitLab chart with UBI-based images

GitLab offers Red Hat UBI
versions of its images, allowing you to replace standard images with UBI-based
images. These images use the same tag as standard images with -ubi8 extension.

The GitLab chart uses third-party images that are not based on UBI. These images
are mostly offer external services to GitLab, such as Redis, PostgreSQL, and so on.
If you wish to deploy a GitLab instance that purely based on UBI you must
disable the internal services, and use external deployments or services.

The services that must be disabled and provided externally are:


  • PostgreSQL
  • MinIO (Object Store)
  • Redis

The services must be disabled are:


  • CertManager (Let’s Encrypt integration)
  • Prometheus
  • Grafana
  • GitLab Runner

Sample values

We provide an example for GitLab chart values in examples/ubi/values.yaml
which can help you to build a pure UBI GitLab deployment.

Known Limitations


  • Currently there is no UBI version of GitLab Runner. Therefore we disable it.
    However, that does not prevent attaching your own runner to your UBI-based
    GitLab deployment.
  • GitLab relies on the official image of Docker Registry which is based on alpine .
    At the moment we do not maintain or release a UBI-based version of Registry. Since
    this functionality is crucial , we do not disable this service.
Read article

Goals | GitLab






  • Scheduler
  • Helm charts

Goals

We have a few core goals with this initiative:


  1. Easy to scale horizontally
  2. Easy to deploy, upgrade, maintain
  3. Wide support of cloud service providers
  4. Initial support for Kubernetes and Helm, with flexibility to support other
    schedulers in the future

Scheduler

We will launch with support for Kubernetes, which is mature and widely supported
across the industry. As part of our design however, we will try to avoid decisions
which will preclude the support of other schedulers. This is especially true for
downstream Kubernetes projects like OpenShift and Tectonic. In the future other
schedulers may also be supported like Docker Swarm and Mesosphere.

We aim to support the scaling and self-healing capabilities of Kubernetes:


  • Readiness and Health checks to ensure pods are functioning, and if not to recycle them
  • Tracks to support canary and rolling deployments
  • Auto-scaling

We will try to leverage standard Kubernetes features:


  • ConfigMaps for managing configuration. These will then get mapped or passed to
    Docker containers
  • Secrets for sensitive data

Since we might be also using Consul, this may be utilized instead for consistency with other installation methods.

Helm charts

A Helm chart will be created to manage the deployment of each GitLab specific container/service. We will then also include bundled charts to make the overall deployment easier. This is particularly
important for this effort, as there will be significantly more complexity in
the Docker and Kubernetes layers than the all-in-one Omnibus based solutions.
Helm can help to manage this complexity, and provide an easy top level interface
to manage settings via the values.yaml file.

We plan to offer a three tiered set of Helm charts:

Helm chart Structure

Read article

Architecture of Cloud native GitLab Helm charts | GitLab





Architecture of Cloud native GitLab Helm charts

Documentation Organization:


  • Goals
  • Architecture
  • Design Decisions
  • Resource Usage
Read article

Resource usage | GitLab







  • Resource Requests

    • GitLab Shell
    • Webservice
    • Sidekiq
    • KAS

Resource usage

Resource Requests

All of our containers include predefined resource request values. By default we
have not put resource limits into place. If your nodes do not have excess memory
capacity, one option is to apply memory limits, though adding more memory (or nodes)
would be preferable. (You want to avoid running out of memory on any of your
Kubernetes nodes, as the Linux kernel’s out of memory manager may end essential Kube processes)

In order to come up with our default request values, we run the application, and
come up with a way to generate various levels of load for each service. We monitor the
service, and make a call on what we think is the best default value.

We will measure:



  • Idle Load - No default should be below these values, but an idle process
    isn’t useful, so typically we will not set a default based on this value.


  • Minimal Load - The values required to do the most basic useful amount of work.
    Typically, for CPU, this will be used as the default, but memory requests come with
    the risk of the Kernel reaping processes, so we will avoid using this as a memory default.


  • Average Loads - What is considered average is highly dependent on the installation,
    for our defaults we will attempt to take a few measurements at a few of what we
    consider reasonable loads. (we will list the loads used). If the service has a pod
    autoscaler, we will typically try to set the scaling target value based on these.
    And also the default memory requests.


  • Stressful Task - Measure the usage of the most stressful task the service
    should perform. (Not necessary under load). When applying resource limits, try and
    set the limit above this and the average load values.


  • Heavy Load - Try and come up with a stress test for the service, then measure
    the resource usage required to do it. We currently don’t use these values for any
    defaults, but users will likely want to set resource limits somewhere between the
    average loads/stress task and this value.

GitLab Shell

Load was tested using a bash loop calling nohup git clone <project> <random-path-name> in order to have some concurrency.
In future tests we will try to include sustained concurrent load, to better match the types of tests we have done for the other services.



  • Idle values

    • 0 tasks, 2 pods

      • cpu: 0
      • memory: 5M

  • Minimal Load

    • 1 tasks (one empty clone), 2 pods

      • cpu: 0
      • memory: 5M

  • Average Loads

    • 5 concurrent clones, 2 pods

      • cpu: 100m
      • memory: 5M
    • 20 concurrent clones, 2 pods

      • cpu: 80m
      • memory: 6M

  • Stressful Task

    • SSH clone the Linux kernel (17MB/s)

      • cpu: 280m
      • memory: 17M
    • SSH push the Linux kernel (2MB/s)

      • cpu: 140m
      • memory: 13M
      • Upload connection speed was likely a factor during our tests

  • Heavy Load

    • 100 concurrent clones, 4 pods

      • cpu: 110m
      • memory: 7M

  • Default Requests

    • cpu: 0 (from minimal load)
    • memory: 6M (from average load)
    • target CPU average: 100m (from average loads)

  • Recommended Limits

    • cpu: > 300m (greater than stress task)
    • memory: > 20M (greater than stress task)

Check the troubleshooting documentation
for details on what might happen if gitlab.gitlab-shell.resources.limits.memory is set too low.

Webservice

Webservice resources were analyzed during testing with the
10k reference architecture.
Notes can be found in the Webservice resources documentation.

Sidekiq

Sidekiq resources were analyzed during testing with the
10k reference architecture.
Notes can be found in the Sidekiq resources documentation.

KAS

Until we learn more about our users need, we expect that our users will be using KAS the following way.



  • Idle values

    • 0 agents connected, 2 pods

      • cpu: 10m
      • memory: 55M

  • Minimal Load :

    • 1 agents connected, 2 pods

      • cpu: 10m
      • memory: 55M

  • Average Load : 1 agent is connected to the cluster.

    • 5 agents connected, 2 pods

      • cpu: 10m
      • memory: 65M

  • Stressful Task :

    • 20 agents connected, 2 pods

      • cpu: 30m
      • memory: 95M

  • Heavy Load :

    • 50 agents connected, 2 pods

      • cpu: 40m
      • memory: 150M

  • Extra Heavy Load :

    • 200 agents connected, 2 pods

      • cpu: 50m
      • memory: 315M

The KAS resources defaults set by this chart are more than enough to handle even the 50 agents scenario.
If you are planning to reach what we consider an Extra Heavy Load , then you should consider tweaking the
default to scale up.



  • Defaults : 2 pods, each with

    • cpu: 100m
    • memory: 100M

For more information on how these numbers were calculated, see the
issue discussion.

Read article