FSx for Lustre errors with No such file or directory

Question

Hello, I am using AWS FSx for Lustre persistent 2 with an S3 bucket as a data repository. I enabled auto import and auto export based on NEW, CHANGED, DELETED. Sometimes, when I try to access a file (let's say it's `/bucket/file`), I get an error saying
`/bucket/file: No such file or directory`.

I can `ls` the file without errors. This missing file error happens randomly and is not always the same file.

I have Log Errors and Log Warnings enabled. On CloudWatch, I did see an like this:
```
{
    "LogLevel": "ERROR",
    "EventType": "DataRepository",
    "FilePath": "/bucket/file",
    "FileSystemID": "",
    "Publisher": "AutoExport",
    "UTCTimestamp": "2022-01-10T22:01:44Z",
    "Message": "Failed to Export file because it is being used by another process."
}
```
But, if I understand correctly, the failure to export shouldn't affect whether or not the file exists on FSx for Lustre, just that S3 may be out of date.

The filesystem also has 75% free space, so I don't think it's a low space issue.

Has anyone seen this error and can help me understand what's going on? Are there any suggested workarounds?

Answer

I understand that you have enabled auto import and auto export based on NEW, CHANGED, DELETED for AWS FSx for Lustre persistent 2 with an S3 bucket as a data repository and sometimes see error "Failed to Export file because it is being used by another process." in cloudwatch logs.

This error occurs when Amazon FSx was unable to export the file because it was being modified by another client on the file system. You can retry the DataRepositoryTask after your workflow has finished writing to the file. I would also like to inform you that this could cause due to factors like,
	
	* How you are accessing the file?
	* What S3 storage class (e.g. S3 Standard, S3 IA, Glacier) is the underlying S3 object in?

To deep dive further, I request you to open a support case with AWS Premium support as the following details are required to understand the cause of the issue.

1. Application name and access command? (e.g. cat)
2. What is the output when you run "ls -lrt" in the directory where this file is present? ( This will give us more information to investigate here)
3. FSX ID / region

FSx for Lustre errors with No such file or directory

Contenuto pertinente