[Lustre-devel] tunefs.lustre before mount

Ben Evans Ben.Evans at terascala.com
Tue Jul 9 14:44:22 PDT 2013

Ah, I was thinking more along the lines of setting the filesystem name, mgs information, peer info, etc., not things like timeouts.

-Ben Evans
From: Dilger, Andreas <andreas.dilger at intel.com>
Sent: Saturday, July 06, 2013 12:17 PM
To: Ben Evans
Cc: lustre-devel at lists.lustre.org
Subject: Re: [Lustre-devel] tunefs.lustre before mount

We're actually trying to get rid of tunefs.lustre for setting config parameters, since it complicates the configuration code, and only has a very limited amount of space.

That said, it should be safe to do repeatedly, though I don't think it has ever been tested.  On the flip side, lctl conf_param does have a limit on the number of parameters that can be set before the configuration has to be reset.

Cheers, Andreas

On 2013-07-05, at 14:52, "Ben Evans" <Ben.Evans at terascala.com<mailto:Ben.Evans at terascala.com>> wrote:

Assuming I can get access to all the "correct" parameters, would there be a reason that performing a tunefs.lustre before each mount would be harmful?  It's the most straightforward way I can think of to force Lustre to have the correct configuration (ib addresses, etc.) at all times.

I don't think it should be an issue, I was just wondering if anyone had run into problems repeatedly running tunefs.lustre (1000s of times against each target)

-Ben Evans
Lustre-devel mailing list
Lustre-devel at lists.lustre.org<mailto:Lustre-devel at lists.lustre.org>

More information about the lustre-devel mailing list