Add a note on a kind issue to README #120
Conversation
yuanchen8911
force-pushed
the
kind-bug
branch
3 times, most recently
from
1b388aa to
6b7b90f
Compare
yuanchen8911
changed the title
README.md
Outdated
| 1. `kind` is installed. See the official documentation [here](https://kind.sigs.k8s.io/docs/user/quick-start/#installation). Note that there is a [known issue with kind](https://kind.sigs.k8s.io/docs/user/known-issues/#pod-errors-due-to-too-many-open-files). You may see the error while trying to tail the log of a running pod in a kind cluster: `failed to create fsnotify watcher: too many open files.` The issue can be resolved by increasing the default values for the configuraton parameters in `/proc/sys/fs/inotify`. | ||
| ```console | ||
| sudo sysctl -w fs.inotify.max_user_watches=<a larger value> | ||
| sudo sysctl -w fs.inotify.max_user_instances=<a large value> | ||
| sudo sysctl -w fs.inotify.max_queued_events=<a large value> | ||
| ``` | ||
|
|
There was a problem hiding this comment.
Let's leave out the solution, and just include the link (the solution is well documented in the link itself).
Can we also move this note to after all of the instructions for installing the prerequisites. I feel like it distracts from the flow to have thid detail here, and it can be done after kind is installed (it doesn't have to be done together with it).
There was a problem hiding this comment.
I've moved it after the kubectl log command to put it in the context.
There was a problem hiding this comment.
I've applied the suggested change. I put it here because line 133 (right before this) is a command showing the pod log, which provides the right context. I'm okay with moving it to the beginning of this section too.
There was a problem hiding this comment.
Does it work to update fs.inotify.max_user_watchesafter the kind cluster has been started? If so, then let's move this to line 106, i.e. before the log calls are made (so people see the warning before things suddenly aren't working).
If you have to update fs.inotify.max_user_watches before the kind cluster is started, then I still think my original suggestion is the right place to put this (i.e. before the kind cluster is started).
There was a problem hiding this comment.
"Does it work to update fs.inotify.max_user_watchesafter the kind cluster has been started?"
Yes, that's what I did. There's no need to recreate the kind cluster. How about moving it to line 98 (before the kubectl commands)? It reads like:
Run the examples by following the steps in the demo script
Finally, you can run the various examples contained in the demo/specs/quickstart folder. The README in that directory shows the full script of the demo you can walk through.
cat demo/specs/quickstart/README.mdNote: there is a known issue with kind. You may see an error while trying to tail the log of a running pod in the kind cluster: failed to create fsnotify watcher: too many open files. The issue may be resolved by increasing the value for fs.inotify.max_user_watches.
yuanchen8911
force-pushed
the
kind-bug
branch
2 times, most recently
from
afca8f4 to
dbb4278
Compare
yuanchen8911
force-pushed
the
kind-bug
branch
4 times, most recently
from
5027241 to
a921075
Compare
README.md
Outdated
| Finally, you can run the various examples contained in the `demo/specs/quickstart` folder. | ||
| The `README` in that directory shows the full script of the demo you can walk through. | ||
|
|
||
| Finally, you can run the various examples contained in the `demo/specs/quickstart` folder. The `README` in that directory shows the full script of the demo you can walk through. |
There was a problem hiding this comment.
This doesn't seem to change anything other than pull the second sentence onto the same line as the first (which I'd like to avoid).
README.md
Outdated
| **Note:** there is a [known issue with kind](https://kind.sigs.k8s.io/docs/user/known-issues/#pod-errors-due-to-too-many-open-files). You may see an error while trying to tail the log of a running pod in the kind cluster: `failed to create fsnotify watcher: too many open files.` The issue may be resolved by increasing the value for `fs.inotify.max_user_watches`. | ||
|
|
There was a problem hiding this comment.
I still think this belongs further down at line 106, i.e. just after the sentence:
Get the pods' statuses. Depending on which GPUs are available, running the first three examples will produce output similar to the following.
There was a problem hiding this comment.
OK, it's been moved to line 106 as suggested.
yuanchen8911
force-pushed
the
kind-bug
branch
3 times, most recently
from
b3fdf68 to
9f99f30
Compare
Signed-off-by: Yuan Chen <[email protected]> Fix a format in README Signed-off-by: Yuan Chen <[email protected]> Rremove .DS_Store Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Update README.md Signed-off-by: Yuan Chen <[email protected]> Update README.md Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Update README.md Co-authored-by: Kevin Klues <[email protected]> Signed-off-by: Yuan Chen <[email protected]> Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Update README Signed-off-by: Yuan Chen <[email protected]> Fix a format issue in README Signed-off-by: Yuan Chen <[email protected]> Fix a format issue in README Signed-off-by: Yuan Chen <[email protected]> Fix a format issue in README Signed-off-by: Yuan Chen <[email protected]> Update README.md Co-authored-by: Kevin Klues <[email protected]> Signed-off-by: Yuan Chen <[email protected]> Signed-off-by: Yuan Chen <[email protected]>
This PR adds a note on a kind issue to README.
Checking a pod's log in a kind cluster may report the following error: