Conversation
…device addresses The three driver subdirectories carry the libraries and the topology file but not hccn_tool, and a transport building Device RoCE endpoints reads each NPU's NIC address through it: a container holding the libraries without the tool cannot open that transport at all, and every process that does dies at startup reporting a device-ip lookup. /etc/hccn.conf, added for the same reason, is the other place those addresses can be read from — but only on a host that keeps them there, while the tool answers from the driver on every host that has one. The wider mount covers both cases with one entry and matches what the Kubernetes chart and the worker's own container already mount. The driver belongs to the host on every Ascend containerization path, the image supplying the toolkit, so injecting more of it shadows nothing a container ships with.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request simplifies the Ascend CDI deployment by mounting the entire host driver tree (/usr/local/Ascend/driver) instead of individual subdirectories and the /etc/hccn.conf file. This change ensures that hccn_tool is available to read NPU NIC addresses, preventing initialization failures on hosts where /etc/hccn.conf is absent. Corresponding test assertions have been updated to reflect this change. I have no additional feedback to provide.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The three driver subdirectories carry the libraries and the topology file but not hccn_tool, and a transport building Device RoCE endpoints reads each NPU's NIC address through it: a container holding the libraries without the tool cannot open that transport at all, and every process that does dies at startup reporting a device-ip lookup.
/etc/hccn.conf, added for the same reason, is the other place those addresses can be read from — but only on a host that keeps them there, while the tool answers from the driver on every host that has one. The wider mount covers both cases with one entry and matches what the Kubernetes chart and the worker's own container already mount.
The driver belongs to the host on every Ascend containerization path, the image supplying the toolkit, so injecting more of it shadows nothing a container ships with.