Backup/restore NFS/SMB share config

[John Beranek - PA]

Hi all,

I've been looking around for a mechanism to backup/restore NFS and SMB share config, and as yet haven't found anything useful.

I'm looking to script the backup, removal and reinstatement of share configuration, as part of a DR failover plan.

Neither the 'isi' or 'isi_classic' commands seem capable of exporting or importing these settings...is there some other way?

Cheers,

John

[Dan Pritts]

Take a look at the platform API.  I bet there's a way to download a share config as a json blob, and there must be a way to push the same back up.

John Beranek - PA

December 15, 2014 at 6:33 AM

--
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.
For more options, visit https://groups.google.com/d/optout.

--
Dan Pritts
ICPSR Computing & Network Services
University of Michigan
+1 (734)615-7362

[Keith Nargi]

You might be hard pressed to be able to backup the config for shares and exports and reinstate them as part of DR.  The shares config are part of gconfig and depending on the version of onefs are backed either by a flat file with all the share info pre7.1.1 or a database post 7.1.1.  To script this you need some insight into OneFS's code to do this.  NFS as of 7.2 is now in the user space and will be part of gconfig as well prior to that Isilon used a nfs server that was part of the kernel so you would need to dig into the server to pull all the export configs out.  

Basically it's a very tedious process.  You can work with PS to help with this but fwiw part of the roadmap for 2015 is to have config replication so you might see it as part of the product next year. 

--
Keith 

[John Beranek - PA]

In this case, I'm looking to improve planned failover, so to automate/assist:

1. Removing access to the share(s) to be migrated to the secondary cluster
2. Perform final run of SyncIQ policy
3. Enable writes on the secondary cluster directory
4. Prepare re-sync policy

Was looking to do 1. by deleting exports/shares on the primary cluster during the failover, and then recreating them after the re-sync policy has been built, after the domain mark has done its job.

John

[Andrew MacKay]

Thought I would post here to let you know the config replication issue and SyncIQ automated failover is now available as a product,  My company has launched a product working closely with Isilon product group.  You can read about it here http://superna.net/eyeglass-isilon/  It will be sold through by EMC through EMC Select in April 2015.  In addition it has full automation abilities for Isilon in general using the API and will have client redirection automation with SyncIQ.   You can contact me for a demo or register for free online training course here http://superna.net/eyeglassonlineform/

Andrew Mackay (CTO of Superna)

[Jamie Ivanov]

The cluster-wide configuration is already backed up with every logset that is generated. There is no standard procedure for backup/restoring these files but they are contained within /ifs/.ifsvar. Also keep in mind that this directory contains a lot of instance-specific data so it would not be advised to simply backup/restore this path as that can lead to a wealth of unhappiness. The structure of these files may change with differing OneFS versions and with different patches so if you do locate the NFS and SMB share configuration data, they would need to be restored on the same OneFS release with the same patches installed.

An alternative option is to script the running of specific commands and parsing the output to build your own backup/restore solution until the cluster configuration backup/restore features are finished within OneFS. I am more than willing to assist in this but I have a lot more work to do on my KVM environment to get a OneFS virtual cluster running (though I may have the VM conversion complete).

[Brad Schlicht]

John,

I was able to accomplish this using PowerShell and the OneFS Platform API v7.1.1. I thought it was rather straight forward and works great. My script runs on a nightly basis and backs up all of the relevant cluster configs to CSV files that live at our DR site (also backed up). I then have another script that has the ability to restore all of the config's from the CSV files to our DR cluster individually or all at the same time depending on the need.

-Brad 

[John Beranek - PA]

Thanks Brad for pointing out what is in fact the obvious solution! I'm already using the Platform API for gathering data for statistics, but hadn't considered it for configuration backup/restore. I think I missed it because all of use of the Platform API so far has been strictly read-only, I've not used any of the write functionality in the API.

If you're happy to pass them on I'd appreciate a copy of your script(s), at least for reference. Given my platform/language of choice, if I were to write it it would likely be a Perl script.

Cheers,

John

[John Beranek - PA]

Also, apologies, as Dan said much the same back in December!

John

[John Beranek - PA]

Already discovered a few interesting gotchas for NFS exports at least, which stop you using the exact JSON blob you get querying an export to create an export.

Namely, specifying any of the following fields at all:

conflicting_paths

unresolved_clients

Defining the snapshot field as it is returned, i.e.:

"snapshot" : "-"

Also, defining both "map_all" and "map_root"is faulted. Other than that, managed to create an NFS export from a REST client nice and easily.

Deleting my test export with a DELETE was easy too.

John

[Ozen]

Hello All,

There is similar project on EMC Code repo at Github that targeting this.

https://github.com/obergt/Isilon_Tools

Full EMC Code portfolio is also here: http://emccode.github.io

[John Beranek - PA]

Oh nice, an Open Source repository - very in favour of this. :)

Reminds me about the possibility of releasing some of my web statistics/reporting tools into the wild...not sure how my company would take that though.

John

[John Beranek - PA]

I encountered an error using this script, I've raised an issue in the Github:

https://github.com/obergt/Isilon_Tools/issues/1

Cheers,

John

On Friday, 20 March 2015 17:56:13 UTC, Ozen wrote:

[Eric Peskin]

Do these scripts work with OneFS 7.2 ?

Specifically, I'd be saving from a cluster running 7.2.0.2 and restoring to a different cluster running 7.2.0.3.
Thanks,

Eric