| Message ID | 20250127215019.352509-1-anna@kernel.org (mailing list archive) |
|---|---|
| Headers | show |
| Series | NFS & SUNRPC: Sysfs Improvements | expand |
On 27 Jan 2025, at 16:50, Anna Schumaker wrote: > From: Anna Schumaker <anna.schumaker@oracle.com> > > These are a handful of improvements that have been in the back of my > mind for a while. The first patch adds a (read-only) file to the NFS > client's sysfs collection to check the server's implementation id. The > remaining patches are on the sunrpc side. > > I did look into the 'nconnect' and 'max_connect' NFS client mount > options and how they interact with adding a new xprt in patch 4. My > reading is that 'nconnect' is just the initial number of connections > made by the NFS client during mounting. That shouldn't disallow adding a > new connection. The 'max_connect' parameter refers to the maximum number > of unique IP addresses in an xprt switch. The new connection I generate > is a clone of the main xprt, not a new address, so this doesn't come > into play either. So I'm convinced adding a new xprt is okay to do here. > > Thoughts? > Anna Looks great, one minor comment on 3/5. For series: Reviewed-by: Benjamin Coddington <bcodding@redhat.com> Ben
From: Anna Schumaker <anna.schumaker@oracle.com> These are a handful of improvements that have been in the back of my mind for a while. The first patch adds a (read-only) file to the NFS client's sysfs collection to check the server's implementation id. The remaining patches are on the sunrpc side. I did look into the 'nconnect' and 'max_connect' NFS client mount options and how they interact with adding a new xprt in patch 4. My reading is that 'nconnect' is just the initial number of connections made by the NFS client during mounting. That shouldn't disallow adding a new connection. The 'max_connect' parameter refers to the maximum number of unique IP addresses in an xprt switch. The new connection I generate is a clone of the main xprt, not a new address, so this doesn't come into play either. So I'm convinced adding a new xprt is okay to do here. Thoughts? Anna Changes for v2: - Added a file for rpc_clnt information - Added a file for on-step xprt deletion Anna Schumaker (5): NFS: Add implid to sysfs sunrpc: Add a sysfs attr for xprtsec sunrpc: Add a sysfs files for rpc_clnt information sunrpc: Add a sysfs file for adding a new xprt sunrpc: Add a sysfs file for one-step xprt deletion fs/nfs/sysfs.c | 60 ++++++++ include/linux/sunrpc/xprtmultipath.h | 1 + net/sunrpc/sysfs.c | 202 +++++++++++++++++++++++++++ net/sunrpc/xprtmultipath.c | 21 +++ 4 files changed, 284 insertions(+)