For loosely coupled pNFS/flexfiles systems, there is often no advantage
at all in going through the MDS for I/O, since the MDS is subject to
the same limitations as all other clients when talking to DSes. If a
DS is unresponsive, I/O through the MDS will fail.
For such systems, the only scalable solution is to have the pNFS clients
retry doing pNFS, and so the protocol now provides a flag that allows
the pNFS server to signal this.
If LAYOUTGET returns FF_FLAGS_NO_IO_THRU_MDS, then we should assume that
the MDS wants the client to retry using these devices, even if they were
previously marked as being unavailable. To do so, we add a helper,
ff_layout_mark_devices_valid() that will be called from layoutget.
Signed-off-by: Trond Myklebust <trond.myklebust@xxxxxxxxxxxxxxx>
---
fs/nfs/flexfilelayout/flexfilelayout.c | 18 ++++++++++++++++--
fs/nfs/flexfilelayout/flexfilelayout.h | 7 +++++++
2 files changed, 23 insertions(+), 2 deletions(-)
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.c b/fs/nfs/flexfilelayout/flexfilelayout.c
index 7fc14b90886e..03516c80855a 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.c
+++ b/fs/nfs/flexfilelayout/flexfilelayout.c
@@ -339,6 +339,19 @@ static void ff_layout_sort_mirrors(struct nfs4_ff_layout_segment *fls)
}
}
+static void ff_layout_mark_devices_valid(struct nfs4_ff_layout_segment *fls)
+{
+ struct nfs4_deviceid_node *node;
+ int i;
+
+ if (!(fls->flags & FF_FLAGS_NO_IO_THRU_MDS))
+ return;
+ for (i = 0; i < fls->mirror_array_cnt; i++) {
+ node = &fls->mirror_array[i]->mirror_ds->id_node;
+ clear_bit(NFS_DEVICEID_UNAVAILABLE, &node->flags);
+ }
+}
+
static struct pnfs_layout_segment *
ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
struct nfs4_layoutget_res *lgr,
@@ -499,6 +512,7 @@ ff_layout_alloc_lseg(struct pnfs_layout_hdr *lh,
rc = ff_layout_check_layout(lgr);
if (rc)
goto out_err_free;
+ ff_layout_mark_devices_valid(fls);
ret = &fls->generic_hdr;
dprintk("<-- %s (success)\n", __func__);
@@ -1035,7 +1049,8 @@ static int ff_layout_async_handle_error_v4(struct rpc_task *task,
rpc_wake_up(&tbl->slot_tbl_waitq);
/* fall through */
default:
- if (ff_layout_has_available_ds(lseg))
+ if (ff_layout_no_fallback_to_mds(lseg) ||
+ ff_layout_has_available_ds(lseg))
return -NFS4ERR_RESET_TO_PNFS;
reset:
dprintk("%s Retry through MDS. Error %d\n", __func__,
@@ -1153,7 +1168,6 @@ static void ff_layout_io_track_ds_error(struct pnfs_layout_segment *lseg,
}
/* NFS_PROTO call done callback routines */
-
static int ff_layout_read_done_cb(struct rpc_task *task,
struct nfs_pgio_header *hdr)
{
diff --git a/fs/nfs/flexfilelayout/flexfilelayout.h b/fs/nfs/flexfilelayout/flexfilelayout.h
index 68cc0d9828f9..2bb08bc6aaf0 100644
--- a/fs/nfs/flexfilelayout/flexfilelayout.h
+++ b/fs/nfs/flexfilelayout/flexfilelayout.h
@@ -10,6 +10,7 @@
#define FS_NFS_NFS4FLEXFILELAYOUT_H
#define FF_FLAGS_NO_LAYOUTCOMMIT 1
+#define FF_FLAGS_NO_IO_THRU_MDS 2
#include "../pnfs.h"
@@ -146,6 +147,12 @@ FF_LAYOUT_MIRROR_COUNT(struct pnfs_layout_segment *lseg)
}
static inline bool
+ff_layout_no_fallback_to_mds(struct pnfs_layout_segment *lseg)
+{
+ return FF_LAYOUT_LSEG(lseg)->flags & FF_FLAGS_NO_IO_THRU_MDS;
+}
+
+static inline bool
ff_layout_test_devid_unavailable(struct nfs4_deviceid_node *node)
{
return nfs4_test_deviceid_unavailable(node);
--
2.5.0
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
