isi_vol_copy fails if destination is not empy

[Paul Letta]

Hello,
So I’ve been testing out using isi_vol_copy with a  very simple file system.   I’ve found that if the destination directory has anything in it all… the isi_vol_copy fails.

# isi_vol_copy  xxx:/vol/testvol /ifs/group/test -sa username:password -incr
Migration 12: Loaded Checkpoint 1
snapshot create failed: Operation not permitted

If I empty /ifs/group/test.. it works great.

So the question here is how can incrementals  (-incr)  work if it fails if anything (even a single, zero byte file) is in the destination ?

This is OneFS 9.1

Thanks.
Paul

[Erik Weiman]

I’m pretty sure that isi_vol_copy was considered deprecated with the 8.2 release so I’m not sure what options you’ll have to be able to use it in 9.x. 

--

Erik Weiman 

Sent from my iPhone 7

On Mar 5, 2021, at 4:25 PM, Paul Letta <pcl...@gmail.com> wrote:



--
You received this message because you are subscribed to the Google Groups "Isilon Technical User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isilon-user-gr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isilon-user-group/CAAfag84V1aA8RrVAY%2BA8Nwed9G4oZAtSZQZzPDTNytNqHtQyLg%40mail.gmail.com.

[Anubhav Ahuja]

Check if NDMP is running. It will screw up the snapshot and thats why incremental fails 

--

You received this message because you are subscribed to the Google Groups "Isilon Technical User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isilon-user-gr...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isilon-user-group/CAAfag84V1aA8RrVAY%2BA8Nwed9G4oZAtSZQZzPDTNytNqHtQyLg%40mail.gmail.com.

--

Thanks and Regards
Anubhav Ahuja

[Paul Letta]

Hi, and thanks for the responses.    Yeah, Dell support will not even talk to me about isi_vol_copy... it was almost the quickest "that's not supported" i've ever seen.

It's completely broken in 9.1 if there is anything in the destination dir at all.. even a single 0 byte file will make it fail.

It's the same if NDMP is enabled or disabled on the cluster.   I'll give up it and move on.

Thanks.

To view this discussion on the web visit https://groups.google.com/d/msgid/isilon-user-group/CAKR0uVvZ47PqGa6U_UQm0HkpwmftCAXEuBWh2Nm59NtpD8Y86w%40mail.gmail.com.

[pierluigi...@gmail.com]

I have used isi_vol_copy_vnx ( which is almost the same stuff)   quite a lot in 8.1.2.

It will fail if on the first run the destination directory is not empty. If I remember correctly it fails if the directory exist too.

But if it run on first run ( with the -full parameter ) and you don't touch nothing at all in the destination directory, not evena single byte, then ir will work with the -incr parameter.

My task was to move almost 1PB from a vnx to an isilon.

The VNX was in use and there were no chance to get a robocopy to work, due to several ( customer-side ) reason.

I ended up by using isi_vol_copy_vnx.

Not fast, but for most of the vnx filesystem went very well.

In case of a single problem, either isilon than vnx side, then the whole dir ( command ? ) was to restart from zero, by removing the directory on Onefs and all the snapshot created by the command.

The strategy has been to "divide" the vnx filesystems in several command to launch, so, in case of an error ( sometimes on the vnx the snapshot command failed, for example ) the amount of data to resend were limited.

It took a while for the 1Pb data, but I managed to get the task done.

In case you need help ( not techincal but as user of the command ) feel free to ask.

Pierluigi

[Anubhav Ahuja]

I was using the incorrect server_export # that why I was getting the error. The File System was on Server_3 and I was using the IP address of server_2. After changing the IP, it is working again. The migration is still going on. Hopefully it finishes without an error.