diff mbox

[2/3] pNFS: Don't send COMMITs to the DSes if the server invalidated our layout

Message ID 20170429184845.58057-3-trond.myklebust@primarydata.com (mailing list archive)
State New, archived
Headers show

Commit Message

Trond Myklebust April 29, 2017, 6:48 p.m. UTC
If the layout was invalidated, then assume we should requeue all the
pending writes for the DS in question.

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

Patch

diff --git a/fs/nfs/pnfs_nfs.c b/fs/nfs/pnfs_nfs.c
index dc4111d08a27..7697ac0ff81a 100644
--- a/fs/nfs/pnfs_nfs.c
+++ b/fs/nfs/pnfs_nfs.c
@@ -217,6 +217,13 @@  pnfs_generic_alloc_ds_commits(struct nfs_commit_info *cinfo,
 	for (i = 0; i < fl_cinfo->nbuckets; i++, bucket++) {
 		if (list_empty(&bucket->committing))
 			continue;
+		/*
+		 * If the layout segment is invalid, then let
+		 * pnfs_generic_retry_commit() clean up the bucket.
+		 */
+		if (!pnfs_is_valid_lseg(bucket->clseg) &&
+		    !test_bit(NFS_LSEG_LAYOUTRETURN, &bucket->clseg->pls_flags))
+			continue;
 		data = nfs_commitdata_alloc(false);
 		if (!data)
 			break;