Ellis, I am not aware of a tool to run lfs commands without having the file system mounted. A lot of those operations (maybe all) are handled via ioctls that expect to operate on a file.
A few times on smaller file systems, I have mounted them on one of the lustre servers temporarily without any ill effects. This is not likely something you would want to have running in production all the time, but if you are just looking for a quick way to run a few "lfs setstripe" or other types of administrative commands then it should be fine. There used to be a time when that sort of thing was a big no-no, but I think Lustre has become more resilient in recent versions. I would just avoid doing any heavy IO (like benchmarking) using the servers as clients. -Rick On 7/13/21, 3:48 PM, "lustre-discuss on behalf of Ellis Wilson via lustre-discuss" <[email protected] on behalf of [email protected]> wrote: Hi Lustre folks, A few questions about around configuring file layouts, specifically progressive file layouts: 1. In a freshly stood-up Lustre cluster, if there are no clients yet mounted, are there any Lustre utilities (I’ve not found one) that allows one to perform the equivalent of “lfs setstripe” without an active mount point (say, from the MDS node)? 2. If not, is there a reasonable API against which such a utility could be constructed, or is this request at odds with the architecture? 3. In the absence of a separate client to mount the filesystem to perform normal “lfs” commands, can one safely mount the cluster directly from an MDS or some other node within the Lustre FS proper? My understanding is that it is not safe, but that’s based on hearsay, so I’d love to get a more authoritative answer. Thanks to anybody who can help answer one or more of these! ellis _______________________________________________ lustre-discuss mailing list [email protected] http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
