-
Notifications
You must be signed in to change notification settings - Fork 29
Conversation
Return object, not array of single object
In server callback have node send over the sha1 of the file contents - not the whole file itself
Faster than previous working version as we sending file data across the wire
Add sample jsdoc documentation, remove logs
batnode.js
Outdated
} | ||
} | ||
}) | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I agree that the auditShardData
contains a lot of logic and it will be good if we can move part of the logic elsewhere(like the example in your comment)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, it took a little while but I agreed and I did that. See the latest commit.
One thing that it seems like I should also do now that I think of it is make use of the more generalized
and this both didn't work. Am I not using your
Most recent commit has this code, and it breaks audit. Use previous commit if you want to test it out when it works. |
Looks good for the changes. |
@dylankb It looks like you're still using I know that there was one method you were using in which the callback passed to |
Yeah, I had just reverted back to the working code so maybe the branch UI hadn't refreshed because I no longer am trying to use Yes, I agree that what you're saying is the goal. We did try and do more less that in the live on our call, but for some reason it caused problems so I'm leaving a note in this issue to follow up on it later. |
@WilfredTA Let me know if you want to give this a 👍 or if you want me to make any additional changes. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK to merge
Steps to Test:
audit/client/node.js
for uploading a fileserver
, andserver2
nodes up in separate terminals.Notes
Naive auditing. Auditing currently checks that every node holding a redundant shard passes a data integrity check. This is a bit naive, but we need to decide if we want to do something else.
Slow checks? The auditing check code that transforms the manifest and checks each shard duplicate seemed to slow running an audit down a bit.
auditShardData
has a lot of logic in it, so I was thinking of trying to bring some of it up higher, maybe like this, but interested in what you all think.