You are browsing a read-only backup copy of Wikitech. The live site can be found at wikitech.wikimedia.org

Portal:Toolforge/Admin/Kubernetes/Certificates: Difference between revisions

From Wikitech-static
Jump to navigation Jump to search
imported>Arturo Borrero Gonzalez
(→‎operations: mention where the clush command should go in the workflow)
imported>David Caro
 
(One intermediate revision by one other user not shown)
Line 1: Line 1:
This page contains information on '''certificates''' (PKI, X.509, etc) for the '''Toolforge Kubernetes''' cluster.
This page contains information on '''certificates''' (PKI, X.509, etc) for the '''Toolforge Kubernetes''' cluster.


= General considerations =
== General considerations ==


[[File:Toolforge_K8s_PKI_Design_in_Simple_Form.png|right|thumb]]
[[File:Toolforge_K8s_PKI_Design_in_Simple_Form.png|right|thumb]]
Line 15: Line 15:
Most certs can be checked for expiration with <code>sudo kubeadm alpha certs check-expiration</code> on a control plane node.
Most certs can be checked for expiration with <code>sudo kubeadm alpha certs check-expiration</code> on a control plane node.


= Use cases and operations =
== External API access ==
 
Description of the different certificate types we have in the cluster.
 
== external API access ==


We have certain entities contacting external the kubernetes API. The authorization/authentication access is managed using a kubernetes ServiceAccount and a x509 certificate.
We have certain entities contacting external the kubernetes API. The authorization/authentication access is managed using a kubernetes ServiceAccount and a x509 certificate.
Line 28: Line 24:
* '''TODO:''' any other example?
* '''TODO:''' any other example?


=== operations ===
=== Operations ===
{{warning|disable puppet fleetwide to make this whole operation more atomic, and no puppet client see the private repo without content}}
{{warning|disable puppet fleetwide to make this whole operation more atomic, and no puppet client see the private repo without content}}


Line 40: Line 36:


<syntaxhighlight lang="shell-session">
<syntaxhighlight lang="shell-session">
user@tools-clushmaster-02:~$ clush WHATEVER_DISABLE_PUPPET_FEELWIDE # TODO
root@cloud-cumin-03:~# cumin "O{project:tools} AND O{name:tools-prometheus}" 'puppet agent --disable "T12345 refreshing certificates"'


user@tools-k8s-control-3:~$ sudo -i wmcs-k8s-get-cert prometheus
user@tools-k8s-control-3:~$ sudo -i wmcs-k8s-get-cert prometheus
Line 54: Line 50:
-----END RSA PRIVATE KEY-----
-----END RSA PRIVATE KEY-----


root@tools-puppetmaster-02:/var/lib/git/labs/private# stg uncommit -t a706eb28
root@tools-puppetmaster-02:/var/lib/git/labs/private# vim modules/secret/secrets/ssl/toolforge-k8s-prometheus.key
uncommit the patch that modifies 'modules/secret/secrets/ssl/toolforge-k8s-prometheus.key'
# copy paste here the private key
root@tools-puppetmaster-02:/var/lib/git/labs/private# stg pop ; stg push
root@tools-puppetmaster-02:/var/lib/git/labs/private# git commit -a
until you are in the right uncommited patch
# Write the task you are working on in the commit and any details you find relevant
root@tools-puppetmaster-02:/var/lib/git/labs/private# nano modules/secret/secrets/ssl/toolforge-k8s-prometheus.key ; stg refresh
copy paste here the private key
root@tools-puppetmaster-02:/var/lib/git/labs/private# stg push -a ; stg commit -a
you are done!
you are done!


user@laptop:~/git/wmf/operations/puppet$ nano files/ssl/toolforge-k8s-prometheus.crt
user@laptop:~/git/wmf/operations/puppet$ vim files/ssl/toolforge-k8s-prometheus.crt
create a patch similar to https://gerrit.wikimedia.org/r/#/c/601692/
create a patch similar to https://gerrit.wikimedia.org/r/#/c/601692/


user@tools-clushmaster-02:~$ clush WHATEVER_ENABLE_PUPPET_FEELWIDE # TODO
root@cloud-cumin-03:~# cumin "O{project:tools} AND O{name:tools-prometheus}" 'puppet agent --enable'
[...]
</syntaxhighlight>
</syntaxhighlight>


== internal API access ==
== Internal API access ==


Some stuff running inside the kubernetes cluster also require a certificate to access the API server and use a ServiceAccount.
Some stuff running inside the kubernetes cluster also require a certificate to access the API server and use a ServiceAccount.
Line 80: Line 72:
* the internal metrics server (i.e, what ''kubectl top'' uses)
* the internal metrics server (i.e, what ''kubectl top'' uses)


=== operations ===
=== Operations ===


Certificates for this use case can be generated using a custom script we have: [[Portal:Toolforge/Admin/Maintenance#wmcs-k8s-secret-for-cert | wmcs-k8s-secret-for-cert]].
Certificates for this use case can be generated using a custom script we have: [[Portal:Toolforge/Admin/Maintenance#wmcs-k8s-secret-for-cert | wmcs-k8s-secret-for-cert]].
Line 110: Line 102:
</syntaxhighlight>
</syntaxhighlight>


== node/kubelet certs ==
== Node/kubelet certs ==


Kubelet has two certs:
Kubelet has two certs:
Line 118: Line 110:
At this time the serving certificate is a self-signed one managed by kubelet, which should not need manual rotation. Proper, CA-signed rotating certs are stabilizing as a feature set in Kubernetes 1.17, and we should probably switch to that for consistency and as a general improvement. The client cert of kubelet is signed by the cluster CA and expires in 1 year.  
At this time the serving certificate is a self-signed one managed by kubelet, which should not need manual rotation. Proper, CA-signed rotating certs are stabilizing as a feature set in Kubernetes 1.17, and we should probably switch to that for consistency and as a general improvement. The client cert of kubelet is signed by the cluster CA and expires in 1 year.  


=== operations ===
=== Operations ===
All such client certs are rotated when upgrading Kubernetes, but they can be manually rotated with kubeadm as well. This should be as easy as running <code>kubeadm alpha certs renew</code> on a control plane node as root.
All such client certs are rotated when upgrading Kubernetes, but they can be manually rotated with kubeadm as well. This should be as easy as running <code>kubeadm alpha certs renew</code> on a control plane node as root.


Line 125: Line 117:
'''TODO:''' elaborate
'''TODO:''' elaborate


== tool certs ==
== Tool certs ==


These certs are automatically generated by the [https://gerrit.wikimedia.org/r/plugins/gitiles/labs/tools/maintain-kubeusers/ maintain-kubeusers] mechanism. When a new tool is created in Striker, the LDAP change is picked up by a polling loop in the maintain-kubeusers deployment, and the service will:
These certs are automatically generated by the [https://gerrit.wikimedia.org/r/plugins/gitiles/labs/tools/maintain-kubeusers/ maintain-kubeusers] mechanism. When a new tool is created in Striker, the LDAP change is picked up by a polling loop in the maintain-kubeusers deployment, and the service will:
Line 141: Line 133:
This service runs in Kubernetes in a specialized namespace just for it using a hand-made Docker image, as is documented in the README of the repo. The toolsbeta version runs the <code>maintain-kubeusers:beta</code> tag instead of the <code>:latest</code> tag to facilitate staging and testing live without hurting Toolforge proper. Deploying new code only requires deleting the currently-running pod after refreshing the required image tag.
This service runs in Kubernetes in a specialized namespace just for it using a hand-made Docker image, as is documented in the README of the repo. The toolsbeta version runs the <code>maintain-kubeusers:beta</code> tag instead of the <code>:latest</code> tag to facilitate staging and testing live without hurting Toolforge proper. Deploying new code only requires deleting the currently-running pod after refreshing the required image tag.


=== operations ===
=== Operations ===


If someone has a need to rotate their tool user certs for some reason, run:
If someone has a need to rotate their tool user certs for some reason, run:
<syntaxhighlight lang=shell-session>user@bastion $ sudo -i
<syntaxhighlight lang=shell-session>user@bastion $ sudo become <tool-that-needs-help>
root@bastion # become <tool-that-needs-help>
tools.toolname:~$ kubectl delete cm maintain-kubeusers
tools.toolname:~$ kubectl delete cm maintain-kubeusers
</syntaxhighlight>
</syntaxhighlight>
This will cause maintain-kubeusers to refresh their certs.
This will cause maintain-kubeusers to refresh their certs.


If the certs are deleted, you will need to instead run <code>kubectl delete cm maintain-kubeusers --namespace tool-$toolname</code> as a cluster admin (such as root on a control plane node) since the tool won't be able to authenticate.
If the certs were already deleted, you will need to instead have a cluster admin run <code>kubectl delete cm maintain-kubeusers --namespace tool-$toolname --as-group=system:masters --as=admin</code> since the tool won't be able to authenticate.


In case of a corrupt <code>.kube/config</code> file, the same trick applies except, that <code>maintain-kubeusers</code> will not read invalid YAML. Therefore, you will need to delete the tool's <code>.kube/config</code> and then as a cluster admin, run <code>kubectl delete cm maintain-kubeusers --namespace tool-$toolname</code>.  That will regenerate their credentials.
In case of a corrupt <code>.kube/config</code> file, the same trick applies except, that <code>maintain-kubeusers</code> will not read invalid YAML. Therefore, you will need to delete the tool's <code>.kube/config</code> and then as a cluster admin, run <code>kubectl delete cm maintain-kubeusers --namespace tool-$toolname --as-group=system:masters --as=admin</code>.  Maintain-kubeusers will regenerate their credentials soon after.


== etcd certs ==
== Etcd certs ==


All etcd servers use puppetmaster issued certificates (puppet node certificates). The etcd service will only allow communication from clients presenting a certificate signed by the same CA.
All etcd servers use puppetmaster issued certificates (puppet node certificates). The etcd service will only allow communication from clients presenting a certificate signed by the same CA.
Line 161: Line 152:
In the puppet profile controlling this, we have a mechanism to refresh the certificate and restart the etcd daemon if the puppet node certificate changes (it is reissued or whatever).
In the puppet profile controlling this, we have a mechanism to refresh the certificate and restart the etcd daemon if the puppet node certificate changes (it is reissued or whatever).


= See also =
== See also ==


Some other interesting docs:
Some other interesting docs:

Latest revision as of 12:38, 2 June 2022

This page contains information on certificates (PKI, X.509, etc) for the Toolforge Kubernetes cluster.

General considerations

Toolforge K8s PKI Design in Simple Form.png

Kubernetes includes an internal CA which is the main one we use for cluster operations.

By default, kubernetes issued certificates are valid for 1 year. After that period, they should be renewed.

The internal kubernetes CA, generated at deployment time by kubadm expires after 10 years. The current CA is good until Nov 3 14:13:50 2029 GMT

Worth noting that etcd servers don't use the kubernetes CA, but use the puppetmaster CA instead.

Most certs can be checked for expiration with sudo kubeadm alpha certs check-expiration on a control plane node.

External API access

We have certain entities contacting external the kubernetes API. The authorization/authentication access is managed using a kubernetes ServiceAccount and a x509 certificate. The x509 certificate encodes the ServiceAccount name in the Subject field.

Some examples of this:

  • tools-prometheus uses this external API access to scrape metrics.
  • TODO: any other example?

Operations

Certificates for this use case can be generated using a custom script we have: wmcs-k8s-get-cert .

Usually, the generated cert will be copy&pasted into the private puppet repo to be used as a secret in a puppet module or profile.

Renewing the certificate is just generating a new one and replacing the old one.

Example workflow for replacing tools-prometheus k8s certificate:

root@cloud-cumin-03:~# cumin "O{project:tools} AND O{name:tools-prometheus}" 'puppet agent --disable "T12345 refreshing certificates"'

user@tools-k8s-control-3:~$ sudo -i wmcs-k8s-get-cert prometheus
/tmp/tmp.9k9N7ksn6K/server-cert.pem
/tmp/tmp.9k9N7ksn6K/server-key.pem
user@tools-k8s-control-3:~$ sudo cat /tmp/tmp.9k9N7ksn6K/server-cert.pem
-----BEGIN CERTIFICATE-----
MIIDYTCCA[...]
-----END CERTIFICATE-----
user@tools-k8s-control-3:~$ sudo cat /tmp/tmp.9k9N7ksn6K/server-key.pem
-----BEGIN RSA PRIVATE KEY-----
MIIEpQIBA[...]
-----END RSA PRIVATE KEY-----

root@tools-puppetmaster-02:/var/lib/git/labs/private# vim modules/secret/secrets/ssl/toolforge-k8s-prometheus.key
# copy paste here the private key
root@tools-puppetmaster-02:/var/lib/git/labs/private# git commit -a
# Write the task you are working on in the commit and any details you find relevant
you are done!

user@laptop:~/git/wmf/operations/puppet$ vim files/ssl/toolforge-k8s-prometheus.crt
create a patch similar to https://gerrit.wikimedia.org/r/#/c/601692/

root@cloud-cumin-03:~# cumin "O{project:tools} AND O{name:tools-prometheus}" 'puppet agent --enable'

Internal API access

Some stuff running inside the kubernetes cluster also require a certificate to access the API server and use a ServiceAccount. This certificate is usually crafted as a Kubernetes secret for the utility to use it.

Some examples of this:

  • our custom webhook: ingress admission controller
  • our custom webhook: registry admission controller
  • the internal metrics server (i.e, what kubectl top uses)

Operations

Certificates for this use case can be generated using a custom script we have: wmcs-k8s-secret-for-cert.

After running the script, the secret should be ready to use.

Renewing the certificate is just generating a new one (running the script again and making sure the pod uses it).

If you want to make sure the old cert is no longer present, just delete it and run the script again. Example session for the metrics-server:

root@tools-k8s-control-3:~# kubectl delete secrets -n metrics metrics-server-certs
secret "metrics-server-certs" deleted
root@tools-k8s-control-3:~# wmcs-k8s-secret-for-cert -n metrics -s metrics-server-certs -a metrics-server
secret/metrics-server-certs created
root@tools-k8s-control-3:~# kubectl get secrets -n metrics metrics-server-certs -o yaml | grep cert.pem | head -1 | awk -F' ' '{print $2}' | base64 -d | openssl x509 -text
Certificate:
    Data:
        Version: 3 (0x2)
        Serial Number:
            2f:65:a6:cf:2c:16:2f:39:6e:29:95:ee:35:01:b9:d7:75:a1:d2:50
        Signature Algorithm: sha256WithRSAEncryption
        Issuer: CN = kubernetes
        Validity
            Not Before: Jun  2 11:31:00 2020 GMT
            Not After : Jun  2 11:31:00 2021 GMT
        Subject: CN = metrics-server
[..]

Node/kubelet certs

Kubelet has two certs:

  1. A client cert to communicate with the API server
  2. A serving certificate for the Kubelet API

At this time the serving certificate is a self-signed one managed by kubelet, which should not need manual rotation. Proper, CA-signed rotating certs are stabilizing as a feature set in Kubernetes 1.17, and we should probably switch to that for consistency and as a general improvement. The client cert of kubelet is signed by the cluster CA and expires in 1 year.

Operations

All such client certs are rotated when upgrading Kubernetes, but they can be manually rotated with kubeadm as well. This should be as easy as running kubeadm alpha certs renew on a control plane node as root.

It is possible to configure the kubelet to request upgraded certs on its own when they near expiration. So far, we have not set this flag in the config, expecting our upgrade cycle to be 6 months, roughly.

TODO: elaborate

Tool certs

These certs are automatically generated by the maintain-kubeusers mechanism. When a new tool is created in Striker, the LDAP change is picked up by a polling loop in the maintain-kubeusers deployment, and the service will:

  • Create the NFS folder for the tool if it isn't already there because of maintain-dbusers
  • Create the necessary folders to set up the KUBECONFIG for the user.
  • Create a tool namespace along with all necessary privileges, restrictions and quotas
  • Generate a private key
  • Request and approve the CSR for the cert to authenticate the new tool with the Kubernetes cluster
  • Write out the cert to the appropriate files along with the KUBECONFIG
  • Create a configmap named maintain-kubeusers in the tool namespace that gives the expiration date of the cert to use for automatically regenerating the cert before it expires
    • Deleting this configmap will cause the cert to be regenerated on the next iteration. This is the safest way to regenerate the certs manually.

Each cert includes a CN, which functions as the user name in Kubernetes, and can include groups as well ("O:" or organization entries). Tool certs currently have the CN of their tool name and one O of "toolforge".

This service runs in Kubernetes in a specialized namespace just for it using a hand-made Docker image, as is documented in the README of the repo. The toolsbeta version runs the maintain-kubeusers:beta tag instead of the :latest tag to facilitate staging and testing live without hurting Toolforge proper. Deploying new code only requires deleting the currently-running pod after refreshing the required image tag.

Operations

If someone has a need to rotate their tool user certs for some reason, run:

user@bastion $ sudo become <tool-that-needs-help>
tools.toolname:~$ kubectl delete cm maintain-kubeusers

This will cause maintain-kubeusers to refresh their certs.

If the certs were already deleted, you will need to instead have a cluster admin run kubectl delete cm maintain-kubeusers --namespace tool-$toolname --as-group=system:masters --as=admin since the tool won't be able to authenticate.

In case of a corrupt .kube/config file, the same trick applies except, that maintain-kubeusers will not read invalid YAML. Therefore, you will need to delete the tool's .kube/config and then as a cluster admin, run kubectl delete cm maintain-kubeusers --namespace tool-$toolname --as-group=system:masters --as=admin. Maintain-kubeusers will regenerate their credentials soon after.

Etcd certs

All etcd servers use puppetmaster issued certificates (puppet node certificates). The etcd service will only allow communication from clients presenting a certificate signed by the same CA. This means kubernetes components that contact etcd should use puppet node certificates.

In the puppet profile controlling this, we have a mechanism to refresh the certificate and restart the etcd daemon if the puppet node certificate changes (it is reissued or whatever).

See also

Some other interesting docs: