Fix log rotation description in the logging doc #3918
Conversation
k8s-ci-robot
added
the
cncf-cla: yes
|
/cc @piosz @fgrzadkowski |
| directory are rotated daily and based on the log size. However, | ||
| system component logs have a higher size retention: by default, | ||
| they can store up to 100MB. | ||
| directory should be rotated. In Kubernetes clusters brough up by |
| they can store up to 100MB. | ||
| directory should be rotated. In Kubernetes clusters brough up by | ||
| `kube-up.sh` script, those logs are configured to be rotated by | ||
| `logrotate` tool daily or once the size exceeds 100MB. |
There was a problem hiding this comment.
add "the": "...the logrotate tool..."
| system component logs have a higher size retention: by default, | ||
| they can store up to 100MB. | ||
| directory should be rotated. In Kubernetes clusters brough up by | ||
| `kube-up.sh` script, those logs are configured to be rotated by |
There was a problem hiding this comment.
add "the": "...the kube-up.sh script..."
| When you run [`kubectl logs`](/docs/user-guide/kubectl/v1.6/#logs), as in the basic logging example, the kubelet on the node handles the request and reads directly from the log file, returning the contents in the response. Note that `kubectl logs` **only returns the last rotation**; you must manually extract prior rotations, if desired and cluster-level logging is not enabled. | ||
| An important consideration in node-level logging is implementing log rotation, | ||
| so that logs don't consume all available storage on the node. Kubernetes | ||
| up to 1.7 is not responsible for rotating logs, deployment tool should set it |
There was a problem hiding this comment.
Suggested edit: "Kubernetes versions less than 1.7 are not responsible for rotating logs, so a deployment tool should be set up to address that."
There was a problem hiding this comment.
Do we want to refer to a specific kubelet version? We'll have to update it every release.
There was a problem hiding this comment.
It's still not responsible and it's unknown when/whether it'll be responsible
I'll try to make it clearer
There was a problem hiding this comment.
Do we want to refer to a specific kubelet version? We'll have to update it every release.
When I was addressing this part, I forgot to refresh the page and this comment wasn't visible
I agree, it's better to remove the mention of the version altogether
| An important consideration in node-level logging is implementing log rotation, | ||
| so that logs don't consume all available storage on the node. Kubernetes | ||
| up to 1.7 is not responsible for rotating logs, deployment tool should set it | ||
| up. For example, in Kubernetes clusters, deployed by `kube-up.sh` script, |
There was a problem hiding this comment.
add "the": "...the kube-up.sh script..."
| so that logs don't consume all available storage on the node. Kubernetes | ||
| up to 1.7 is not responsible for rotating logs, deployment tool should set it | ||
| up. For example, in Kubernetes clusters, deployed by `kube-up.sh` script, | ||
| there is [`logrotate`](http://www.linuxcommand.org/man_pages/logrotate8.html) |
There was a problem hiding this comment.
add "a": "...there is a logrotate tool..."
| up to 1.7 is not responsible for rotating logs, deployment tool should set it | ||
| up. For example, in Kubernetes clusters, deployed by `kube-up.sh` script, | ||
| there is [`logrotate`](http://www.linuxcommand.org/man_pages/logrotate8.html) | ||
| tool configured to perform log rotations daily or once application's log |
There was a problem hiding this comment.
add "an": "...or once an application's log..."
| up. For example, in Kubernetes clusters, deployed by `kube-up.sh` script, | ||
| there is [`logrotate`](http://www.linuxcommand.org/man_pages/logrotate8.html) | ||
| tool configured to perform log rotations daily or once application's log | ||
| file grows beyond 10MB. You can also set up container runtime to rotate logs |
There was a problem hiding this comment.
add "a": "You can also set up a container runtime..."
| the basic logging example, the kubelet on the node handles the request and | ||
| reads directly from the log file, returning the contents in the response. | ||
| **Note:** if some external system has manipulated the log file, e.g. | ||
| logrotate performed rotation, only the content of the original log file |
There was a problem hiding this comment.
add "the": "...logrotate performed the rotation..."
|
@chenopis I swear when I'm creating the PR the checkbox "Allow edits from maintainers" is checked, but after PR is created, it's unchecked. That's what I was missing, sorry |
| tool configured to perform log rotations daily or once application's log | ||
| file grows beyond 10MB. You can also set up container runtime to rotate logs | ||
| automatically, e.g. by using Docker's `log-opt` or by sending logs to the | ||
| system that does this for you, e.g. journald. As an example, you can find |
There was a problem hiding this comment.
AFAIU journald doesn't work with kubectl logs or /logs http handler. I wouldn't recommend it in our docs.
| the basic logging example, the kubelet on the node handles the request and | ||
| reads directly from the log file, returning the contents in the response. | ||
| **Note:** if some external system has manipulated the log file, e.g. | ||
| logrotate performed rotation, only the content of the original log file |
There was a problem hiding this comment.
What do you mean by "original"? I think it's unclear.
|
@chenopis @fgrzadkowski Addressed comments, PTAL |
| there is a [`logrotate`](http://www.linuxcommand.org/man_pages/logrotate8.html) | ||
| tool configured to perform log rotations daily or once an application's log | ||
| file grows beyond 10MB. You can also set up a container runtime to rotate logs | ||
| automatically, e.g. by using Docker's `log-opt`. As an example, you can find |
There was a problem hiding this comment.
Aren't we using Docker to rotate logs? I though we now use 'logrotate' only for logs from system components.
There was a problem hiding this comment.
In COS on GCP, yes. Agree, considering also the link below, this part may be confusing. I'll address it, thanks
|
@fgrzadkowski or @piosz does this have tech lgtm? |
|
@chenopis Now it does :) |
…hub.io into release-1.7 * 'master' of https://github.com/kubernetes/kubernetes.github.io: Fix log rotation description in the logging doc (#3918) Add colon to #3897 add prompt about KUBE_REPO_PREFIX (#3597) Warn against managing ReplicaSets owned by Deployments Update small typo in components page (#3930) change the service name to pod ip (#3926)
…hub.io into release-1.6 * 'master' of https://github.com/kubernetes/kubernetes.github.io: networking.md: Add reference to the Cilium network plugin Fix typo in cluster api docs [Kubeadm] bugfix in the index listing Fix log rotation description in the logging doc (#3918) Add colon to #3897 add prompt about KUBE_REPO_PREFIX (#3597) Warn against managing ReplicaSets owned by Deployments Update small typo in components page (#3930) change the service name to pod ip (#3926)
To address kubernetes/kubernetes#40634 (comment)
This change is