The following topics are discussed:
- Upgrading Kubernetes to version 1.6
- Upgrading the Daemon Sets
- CPU and Memory Requirements
- Pod Eviction
- Network Policy Controller
- Troubleshooting Blocked Connections
- Things to watch out for
- Changing Configuration Options
Weave Net can be installed onto your CNI-enabled Kubernetes cluster with a single command:
$ kubectl apply -f "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')"
After a few seconds, a Weave Net pod should be running on each Node and any further pods you create will be automatically attached to the Weave network.
Note: This command requires Kubernetes 1.4 or later, and we recommend your master node has at least two CPU cores.
CNI, the Container Network Interface, is a proposed standard for configuring network interfaces for Linux containers.
If you do not already have a CNI-enabled cluster, you can bootstrap one easily with kubeadm.
Alternatively, you can configure CNI yourself
Note: If using the Weave CNI Plugin from a prior full install of Weave Net with your cluster, you must first uninstall it before applying the Weave-kube addon. Shut down Kubernetes, and on all nodes perform the following:
- Remove any separate provisions you may have made to run Weave at
Then relaunch Kubernetes and install the addon as described above.
In version 1.6, Kubernetes has increased security, so we need to
create a special service account to run Weave Net. This is done in
weave-daemonset-k8s-1.6.yaml attached to the Weave Net
Also, the toleration required to let Weave Net run on master nodes has moved from an annotation to a field on the DaemonSet spec object.
If you have edited the Weave Net DaemonSet from a previous release, you will need to re-make your changes against the new version.
For Kubernetes 1.6 and above the DaemonSet definition specifies Rolling Updates, so when you apply a new version Kubernetes will automatically restart the Weave Net pods one by one.
Kubernetes v1.5 and below does not support rolling upgrades of daemon sets, and so you will need to perform the procedure manually:
- Apply the updated addon manifest
kubectl apply -f "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')"
- Kill each Weave Net pod with
kubectl deleteand then wait for it to reboot before moving on to the next pod.
Note: In versions prior to Weave Net 2.0, deleting all Weave Net pods at the same time will result in them losing track of IP address range ownership, possibly leading to duplicate IP addresses if you then start a new copy of Weave Net.
Kubernetes manages resources on each node, and only schedules pods to run on nodes that have enough free resources.
The components of a typical Kubernetes installation (with the master node running etcd, scheduler, api-server, etc.) take up about 95% of a CPU, which leaves little room to run anything else. For all of Weave Net’s features to work, it must run on every node, including the master.
The best way to resolve this issue is to use machines with at least two CPU cores. However if you are installing Kubernetes and Weave Net for the first time, you may not be aware of this requirement. For this reason, Weave Net launches as a DaemonSet with a specification that reserves at least 1% CPU for each container. This enables Weave Net to start up seamlessly on a single-CPU node.
Depending on the workload, Weave Net may need more than 1% of the CPU. The percentage set in the DaemonSet is the minimum and not a limit. This minimum setting allows Weave Net to take advantage of available CPU and “burst” above that limit if it needs to.
If a node runs out of CPU, memory or disk, Kubernetes may decide to evict one or more pods. It may choose to evict the Weave Net pod, which will disrupt pod network operations.
You can reduce the chance of eviction by changing the DaemonSet to have a much bigger request, and a limit of the same value.
This causes Kubernetes to apply a “guaranteed” rather than a “burstable” policy. However a similar request for disk space can not be made, and so please be aware of this issue and monitor your resources to ensure that they stay below 100%.
You can see when pods have been evicted via the
kubectl get events command
LASTSEEN COUNT NAME KIND TYPE REASON SOURCE MESSAGE 1m 1 mypod-09vkd Pod Warning Evicted kubelet, node-1 The node was low on resource: memory.
kubectl get pods
NAME READY STATUS RESTARTS AGE IP NODE mypod-09vkd 0/1 Evicted 0 1h <none> node-1
If you see this in your cluster, consider some of the above steps to reduce disruption.
The addon also supports the Kubernetes policy API so that you can securely isolate pods from each other based on namespaces and labels. For more information on configuring network policies in Kubernetes see the walkthrough and the NetworkPolicy API object definition.
Note: as of version 1.9 of Weave Net, the Network Policy
Controller allows all multicast traffic. Since a single multicast
address may be used by multiple pods, we cannot implement rules to
isolate them individually. You can turn this behaviour off (block
all multicast traffic) by adding
--allow-mcast=false as an
weave-npc in the YAML configuration.
The status of Weave Net can be checked by running its CLI commands. This can be done in various ways:
1. Install the
weave script and run:
$ weave status Version: 2.0.1 (up to date; next check at 2017/07/10 13:49:29) Service: router Protocol: weave 1..2 Name: 42:8e:e8:c4:52:1b(host-0) Encryption: disabled PeerDiscovery: enabled Targets: 3 Connections: 3 (2 established, 1 failed) Peers: 3 (with 6 established connections) TrustedSubnets: none Service: ipam Status: ready Range: 10.32.0.0/12 DefaultSubnet: 10.32.0.0/12
2. If you don’t want to install additional software onto your hosts, run via
kubectl commands, which produce the exact same outcome as the previous example:
### Identify the Weave Net pods: $ kubectl get pods -n kube-system -l name=weave-net -o wide NAME READY STATUS RESTARTS AGE IP NODE weave-net-1jkl6 2/2 Running 0 1d 10.128.0.4 host-0 weave-net-bskbv 2/2 Running 0 1d 10.128.0.5 host-1 weave-net-m4x1b 2/2 Running 0 1d 10.128.0.6 host-2
The above shows all Weave Net pods available in your cluster. You can see Kubernetes has deployed one Weave Net pod per host, in order to interconnect all hosts.
You then need to:
- choose which pod you want to run your command from (in most cases it doesn’t matter
which one you pick so just pick the first one, e.g. pod
kubectl execto run the
- specify the absolute path
--localbecause it’s running inside a container
$ kubectl exec -n kube-system weave-net-1jkl6 -c weave -- /home/weave/weave --local status Version: 2.0.1 (up to date; next check at 2017/07/10 13:49:29) Service: router Protocol: weave 1..2 Name: 42:8e:e8:c4:52:1b(host-0) Encryption: disabled PeerDiscovery: enabled Targets: 3 Connections: 3 (2 established, 1 failed) Peers: 3 (with 6 established connections) TrustedSubnets: none Service: ipam Status: ready Range: 10.32.0.0/12 DefaultSubnet: 10.32.0.0/12
3. Finally you could also use Weave Cloud and monitor all your pods, including Weave Net’s ones, from there.
For more information see What is Weave Cloud?
If you suspect that legitimate traffic is being blocked by the Weave Network Policy Controller, the first thing to do is check the
weave-npc container’s logs.
To do this, first you have to find the name of the Weave Net pod running on the relevant host:
$ kubectl get pods -n kube-system -o wide | grep weave-net weave-net-08y45 2/2 Running 0 1m 10.128.0.2 host1 weave-net-2zuhy 2/2 Running 0 1m 10.128.0.4 host3 weave-net-oai50 2/2 Running 0 1m 10.128.0.3 host2
Select the relevant container, for example, if you want to look at host2 then pick
weave-net-oai50 and run:
$ kubectl logs <weave-pod-name-as-above> -n kube-system weave-npc
When the Weave Network Policy Controller blocks a connection, it logs the following details about it:
- protocol used,
- source IP and port,
- destination IP and port,
as per the below example:
TCP connection from 10.32.0.7:56648 to 10.32.0.11:80 blocked by Weave NPC. UDP connection from 10.32.0.7:56648 to 10.32.0.11:80 blocked by Weave NPC.
- Don’t turn on
--masquerade-allon kube-proxy: this will change the source address of every pod-to-pod conversation which will make it impossible to correctly enforce network policies that restrict which pods can talk.
- If you do set the
--cluster-cidroption on kube-proxy, make sure it matches the
IPALLOC_RANGEgiven to Weave Net (see below)
You can customise the YAML you get from
cloud.weave.works by passing some of Weave Net’s options, arguments and environment variables as query parameters:
version: Weave Net’s version. Default:
latest, i.e. latest release. N.B.: This only changes the specified version inside the generated YAML file, it does not ensure that the rest of the YAML is compatible with that version. To freeze the YAML version save a copy of the YAML file from the release page and use that copy instead of downloading it each time from
password-secret: name of the Kubernetes secret containing your password. Example:
$ echo "s3cr3tp4ssw0rd" > /var/lib/weave/weave-passwd.txt $ kubectl create secret -n kube-system generic weave-passwd --from-file=/var/lib/weave/weave-passwd.txt $ kubectl apply -f "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')&password-secret=weave-passwd"
known-peers: comma-separated list of hosts. Default: empty.
trusted-subnets: comma-separated list of CIDRs. Default: empty.
disable-npc: boolean (
env.NAME=VALUE: add environment variable
NAMEand set it to
seLinuxOptions.NAME=VALUE: add SELinux option
NAMEand set it to
The list of variables you can set is:
CHECKPOINT_DISABLE- if set to 1, disable checking for new Weave Net versions (default is blank, i.e. check is enabled)
HAIRPIN_MODE- Weave Net defaults to enabling hairpin on the bridge side of the
vethpair for containers attached. If you need to disable hairpin, e.g. your kernel is one of those that can panic if hairpin is enabled, then you can disable it by setting
IPALLOC_RANGE- the range of IP addresses used by Weave Net and the subnet they are placed in (CIDR format; default
EXPECT_NPC- set to 0 to disable Network Policy Controller (default is on)
KUBE_PEERS- list of addresses of peers in the Kubernetes cluster (default is to fetch the list from the api-server)
IPALLOC_INIT- set the initialization mode of the IP Address Manager (defaults to consensus amongst the
WEAVE_EXPOSE_IP- set the IP address used as a gateway from the Weave network to the host network - this is useful if you are configuring the addon as a static pod.
WEAVE_MTU- Weave Net defaults to 1376 bytes, but you can set a smaller size if your underlying network has a tighter limit, or set a larger size for better performance if your network supports jumbo frames - see here for more details.
$ kubectl apply -f "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')&env.WEAVE_MTU=1337"
This command – notice
&env.WEAVE_MTU=1337 at the end of the URL – generates a YAML file containing, among others:
[...] containers: - name: weave [...] env: - name: WEAVE_MTU value: '1337' [...]
Note: The YAML file can also be saved for later use or manual editing by using, for example:
$ curl -fsSLo weave-daemonset.yaml "https://cloud.weave.works/k8s/net?k8s-version=$(kubectl version | base64 | tr -d '\n')"
Manually editing the YAML file
Whether you saved the YAML file served from
cloud.weave.works or downloaded a static YAML file from our releases page, you can manually edit it to suit your needs.
- additional arguments may be supplied to the Weave router process by adding them to the
command:array in the YAML file,
- additional parameters can be set via the environment variables listed above; these can be inserted into the YAML file like this:
containers: - name: weave env: - name: IPALLOC_RANGE value: 10.0.0.0/16