diff mbox series

[4/5] pNFS/files: Fall back to I/O through the MDS on non-fatal layout errors

Message ID 20220514140814.3655-4-trondmy@kernel.org (mailing list archive)
State New, archived
Headers show
Series [1/5] NFS: Memory allocation failures are not server fatal errors | expand

Commit Message

Trond Myklebust May 14, 2022, 2:08 p.m. UTC 
From: Trond Myklebust <trond.myklebust@hammerspace.com>

Only report the error when the server is returning a fatal error, such
as ESTALE, EIO, etc...

Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 fs/nfs/filelayout/filelayout.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
diff mbox series

Patch

diff --git a/fs/nfs/filelayout/filelayout.c b/fs/nfs/filelayout/filelayout.c
index 76deddab0a8f..2b2661582bbe 100644
--- a/fs/nfs/filelayout/filelayout.c
+++ b/fs/nfs/filelayout/filelayout.c
@@ -839,7 +839,12 @@  fl_pnfs_update_layout(struct inode *ino,
 
 	lseg = pnfs_update_layout(ino, ctx, pos, count, iomode, strict_iomode,
 				  gfp_flags);
-	if (IS_ERR_OR_NULL(lseg))
+	if (IS_ERR(lseg)) {
+		/* Fall back to MDS on recoverable errors */
+		if (!nfs_error_is_fatal_on_server(PTR_ERR(lseg)))
+			lseg = NULL;
+		goto out;
+	} else if (!lseg)
 		goto out;
 
 	lo = NFS_I(ino)->layout;