minio service chown process on large dataset

[club79]

Hey all,

We have a large pool on TrueNAS 12 U7 which we access using the S3 (minio) service. The disks had to be moved to a different server due to a hardware issue.

Now, when starting the minio service on the new server, it starts the process of taking ownership of all the files again. I don't think it should need to do the chown process because the username and UID are already the same.

Does anyone know if there is a way to force start minio without having it go through this process? Is there any way to bypass this change ownership process?

There doesn't seem to be any way to monitor it's progress, it's very hard to be left in the dark wondering how long this process will take on a large data set.

Anyone have some suggestions on how to get this service started without waiting forever?

 

[anodos]

Hey all,

We have a large pool on TrueNAS 12 U7 which we access using the S3 (minio) service. The disks had to be moved to a different server due to a hardware issue.

Now, when starting the minio service on the new server, it starts the process of taking ownership of all the files again. I don't think it should need to do the chown process because the username and UID are already the same.

Does anyone know if there is a way to force start minio without having it go through this process? Is there any way to bypass this change ownership process?

There doesn't seem to be any way to monitor it's progress, it's very hard to be left in the dark wondering how long this process will take on a large data set.

Anyone have some suggestions on how to get this service started without waiting forever?

If it's something our middleware is doing, then you can view via midclt call core.get_jobs. Once you find the ID, you can periodically monitor via midclt call core.get_jobs '[["id", "=", <id>]]'.

 

[club79]

Apparently the minio service was not actually starting. When I toggled the switch for the service it would flip back off after a few minutes, I had assumed this was due to the permissions needing to be updated.

Now we have the minio service running and and it is going through updating the file ownership.

I remembered that you can get an idea of overall progress using the minio console. The buckets tab on the console will show you the overall size of the data being managed and number of objects, which increases slowly (I'm assuming as it takes ownership or scans the files). In approx 24 hours it's half way at 150TB. As usual, performance with minio is excellent.

If anyone else is planning to enable minio on a large data-set, you could use the console to monitor progress.