Illumos #4045 write throttle & i/o scheduler performance work

4045 zfs write throttle & i/o scheduler performance work

1. The ZFS i/o scheduler (vdev_queue.c) now divides i/os into 5 classes: sync
read, sync write, async read, async write, and scrub/resilver.  The scheduler
issues a number of concurrent i/os from each class to the device.  Once a class
has been selected, an i/o is selected from this class using either an elevator
algorithem (async, scrub classes) or FIFO (sync classes).  The number of
concurrent async write i/os is tuned dynamically based on i/o load, to achieve
good sync i/o latency when there is not a high load of writes, and good write
throughput when there is.  See the block comment in vdev_queue.c (reproduced
below) for more details.

2. The write throttle (dsl_pool_tempreserve_space() and
txg_constrain_throughput()) is rewritten to produce much more consistent delays
when under constant load.  The new write throttle is based on the amount of
dirty data, rather than guesses about future performance of the system.  When
there is a lot of dirty data, each transaction (e.g. write() syscall) will be
delayed by the same small amount.  This eliminates the "brick wall of wait"
that the old write throttle could hit, causing all transactions to wait several
seconds until the next txg opens.  One of the keys to the new write throttle is
decrementing the amount of dirty data as i/o completes, rather than at the end
of spa_sync().  Note that the write throttle is only applied once the i/o
scheduler is issuing the maximum number of outstanding async writes.  See the
block comments in dsl_pool.c and above dmu_tx_delay() (reproduced below) for
more details.

This diff has several other effects, including:

 * the commonly-tuned global variable zfs_vdev_max_pending has been removed;
use per-class zfs_vdev_*_max_active values or zfs_vdev_max_active instead.

 * the size of each txg (meaning the amount of dirty data written, and thus the
time it takes to write out) is now controlled differently.  There is no longer
an explicit time goal; the primary determinant is amount of dirty data.
Systems that are under light or medium load will now often see that a txg is
always syncing, but the impact to performance (e.g. read latency) is minimal.
Tune zfs_dirty_data_max and zfs_dirty_data_sync to control this.

 * zio_taskq_batch_pct = 75 -- Only use 75% of all CPUs for compression,
checksum, etc.  This improves latency by not allowing these CPU-intensive tasks
to consume all CPU (on machines with at least 4 CPU's; the percentage is
rounded up).

--matt

APPENDIX: problems with the current i/o scheduler

The current ZFS i/o scheduler (vdev_queue.c) is deadline based.  The problem
with this is that if there are always i/os pending, then certain classes of
i/os can see very long delays.

For example, if there are always synchronous reads outstanding, then no async
writes will be serviced until they become "past due".  One symptom of this
situation is that each pass of the txg sync takes at least several seconds
(typically 3 seconds).

If many i/os become "past due" (their deadline is in the past), then we must
service all of these overdue i/os before any new i/os.  This happens when we
enqueue a batch of async writes for the txg sync, with deadlines 2.5 seconds in
the future.  If we can't complete all the i/os in 2.5 seconds (e.g. because
there were always reads pending), then these i/os will become past due.  Now we
must service all the "async" writes (which could be hundreds of megabytes)
before we service any reads, introducing considerable latency to synchronous
i/os (reads or ZIL writes).

Notes on porting to ZFS on Linux:

- zio_t gained new members io_physdone and io_phys_children.  Because
  object caches in the Linux port call the constructor only once at
  allocation time, objects may contain residual data when retrieved
  from the cache. Therefore zio_create() was updated to zero out the two
  new fields.

- vdev_mirror_pending() relied on the depth of the per-vdev pending queue
  (vq->vq_pending_tree) to select the least-busy leaf vdev to read from.
  This tree has been replaced by vq->vq_active_tree which is now used
  for the same purpose.

- vdev_queue_init() used the value of zfs_vdev_max_pending to determine
  the number of vdev I/O buffers to pre-allocate.  That global no longer
  exists, so we instead use the sum of the *_max_active values for each of
  the five I/O classes described above.

- The Illumos implementation of dmu_tx_delay() delays a transaction by
  sleeping in condition variable embedded in the thread
  (curthread->t_delay_cv).  We do not have an equivalent CV to use in
  Linux, so this change replaced the delay logic with a wrapper called
  zfs_sleep_until(). This wrapper could be adopted upstream and in other
  downstream ports to abstract away operating system-specific delay logic.

- These tunables are added as module parameters, and descriptions added
  to the zfs-module-parameters.5 man page.

  spa_asize_inflation
  zfs_deadman_synctime_ms
  zfs_vdev_max_active
  zfs_vdev_async_write_active_min_dirty_percent
  zfs_vdev_async_write_active_max_dirty_percent
  zfs_vdev_async_read_max_active
  zfs_vdev_async_read_min_active
  zfs_vdev_async_write_max_active
  zfs_vdev_async_write_min_active
  zfs_vdev_scrub_max_active
  zfs_vdev_scrub_min_active
  zfs_vdev_sync_read_max_active
  zfs_vdev_sync_read_min_active
  zfs_vdev_sync_write_max_active
  zfs_vdev_sync_write_min_active
  zfs_dirty_data_max_percent
  zfs_delay_min_dirty_percent
  zfs_dirty_data_max_max_percent
  zfs_dirty_data_max
  zfs_dirty_data_max_max
  zfs_dirty_data_sync
  zfs_delay_scale

  The latter four have type unsigned long, whereas they are uint64_t in
  Illumos.  This accommodates Linux's module_param() supported types, but
  means they may overflow on 32-bit architectures.

  The values zfs_dirty_data_max and zfs_dirty_data_max_max are the most
  likely to overflow on 32-bit systems, since they express physical RAM
  sizes in bytes.  In fact, Illumos initializes zfs_dirty_data_max_max to
  2^32 which does overflow. To resolve that, this port instead initializes
  it in arc_init() to 25% of physical RAM, and adds the tunable
  zfs_dirty_data_max_max_percent to override that percentage.  While this
  solution doesn't completely avoid the overflow issue, it should be a
  reasonable default for most systems, and the minority of affected
  systems can work around the issue by overriding the defaults.

- Fixed reversed logic in comment above zfs_delay_scale declaration.

- Clarified comments in vdev_queue.c regarding when per-queue minimums take
  effect.

- Replaced dmu_tx_write_limit in the dmu_tx kstat file
  with dmu_tx_dirty_delay and dmu_tx_dirty_over_max.  The first counts
  how many times a transaction has been delayed because the pool dirty
  data has exceeded zfs_delay_min_dirty_percent.  The latter counts how
  many times the pool dirty data has exceeded zfs_dirty_data_max (which
  we expect to never happen).

- The original patch would have regressed the bug fixed in
  zfsonlinux/zfs@c418410, which prevented users from setting the
  zfs_vdev_aggregation_limit tuning larger than SPA_MAXBLOCKSIZE.
  A similar fix is added to vdev_queue_aggregate().

- In vdev_queue_io_to_issue(), dynamically allocate 'zio_t search' on the
  heap instead of the stack.  In Linux we can't afford such large
  structures on the stack.

Reviewed by: George Wilson <[email protected]>
Reviewed by: Adam Leventhal <[email protected]>
Reviewed by: Christopher Siden <[email protected]>
Reviewed by: Ned Bass <[email protected]>
Reviewed by: Brendan Gregg <[email protected]>
Approved by: Robert Mustacchi <[email protected]>

References:
  http://www.illumos.org/issues/4045
  illumos/illumos-gate@69962b5647e4a8b9b14998733b765925381b727e

Ported-by: Ned Bass <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #1913

1 files changed, 21 insertions, 16 deletions

diff --git a/module/zfs/dmu.c b/module/zfs/dmu.c
index 72fce35c5..ade13b9f0 100644
--- a/module/zfs/dmu.c
+++ b/module/zfs/dmu.c
@@ -370,13 +370,11 @@ static int
 dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length,
     int read, void *tag, int *numbufsp, dmu_buf_t ***dbpp, uint32_t flags)
 {
-	dsl_pool_t *dp = NULL;
 	dmu_buf_t **dbp;
 	uint64_t blkid, nblks, i;
 	uint32_t dbuf_flags;
 	int err;
 	zio_t *zio;
-	hrtime_t start = 0;
 
 	ASSERT(length <= DMU_MAX_ACCESS);
 
@@ -404,9 +402,6 @@ dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length,
 	}
 	dbp = kmem_zalloc(sizeof (dmu_buf_t *) * nblks, KM_PUSHPAGE | KM_NODEBUG);
 
-	if (dn->dn_objset->os_dsl_dataset)
-		dp = dn->dn_objset->os_dsl_dataset->ds_dir->dd_pool;
-	start = gethrtime();
 	zio = zio_root(dn->dn_objset->os_spa, NULL, NULL, ZIO_FLAG_CANFAIL);
 	blkid = dbuf_whichblock(dn, offset);
 	for (i = 0; i < nblks; i++) {
@@ -427,9 +422,6 @@ dmu_buf_hold_array_by_dnode(dnode_t *dn, uint64_t offset, uint64_t length,
 
 	/* wait for async i/o */
 	err = zio_wait(zio);
-	/* track read overhead when we are in sync context */
-	if (dp && dsl_pool_sync_context(dp))
-		dp->dp_read_overhead += gethrtime() - start;
 	if (err) {
 		dmu_buf_rele_array(dbp, nblks, tag);
 		return (err);
@@ -511,12 +503,22 @@ dmu_buf_rele_array(dmu_buf_t **dbp_fake, int numbufs, void *tag)
 	kmem_free(dbp, sizeof (dmu_buf_t *) * numbufs);
 }
 
+/*
+ * Issue prefetch i/os for the given blocks.
+ *
+ * Note: The assumption is that we *know* these blocks will be needed
+ * almost immediately.  Therefore, the prefetch i/os will be issued at
+ * ZIO_PRIORITY_SYNC_READ
+ *
+ * Note: indirect blocks and other metadata will be read synchronously,
+ * causing this function to block if they are not already cached.
+ */
 void
 dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, uint64_t len)
 {
 	dnode_t *dn;
 	uint64_t blkid;
-	int nblks, i, err;
+	int nblks, err;
 
 	if (zfs_prefetch_disable)
 		return;
@@ -529,7 +531,7 @@ dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, uint64_t len)
 
 		rw_enter(&dn->dn_struct_rwlock, RW_READER);
 		blkid = dbuf_whichblock(dn, object * sizeof (dnode_phys_t));
-		dbuf_prefetch(dn, blkid);
+		dbuf_prefetch(dn, blkid, ZIO_PRIORITY_SYNC_READ);
 		rw_exit(&dn->dn_struct_rwlock);
 		return;
 	}
@@ -546,16 +548,18 @@ dmu_prefetch(objset_t *os, uint64_t object, uint64_t offset, uint64_t len)
 	rw_enter(&dn->dn_struct_rwlock, RW_READER);
 	if (dn->dn_datablkshift) {
 		int blkshift = dn->dn_datablkshift;
-		nblks = (P2ROUNDUP(offset+len, 1<<blkshift) -
-		    P2ALIGN(offset, 1<<blkshift)) >> blkshift;
+		nblks = (P2ROUNDUP(offset + len, 1 << blkshift) -
+		    P2ALIGN(offset, 1 << blkshift)) >> blkshift;
 	} else {
 		nblks = (offset < dn->dn_datablksz);
 	}
 
 	if (nblks != 0) {
+		int i;
+
 		blkid = dbuf_whichblock(dn, offset);
 		for (i = 0; i < nblks; i++)
-			dbuf_prefetch(dn, blkid+i);
+			dbuf_prefetch(dn, blkid + i, ZIO_PRIORITY_SYNC_READ);
 	}
 
 	rw_exit(&dn->dn_struct_rwlock);
@@ -1559,7 +1563,7 @@ dmu_sync_late_arrival(zio_t *pio, objset_t *os, dmu_sync_cb_t *done, zgd_t *zgd,
 
 	zio_nowait(zio_write(pio, os->os_spa, dmu_tx_get_txg(tx), zgd->zgd_bp,
 	    zgd->zgd_db->db_data, zgd->zgd_db->db_size, zp,
-	    dmu_sync_late_arrival_ready, dmu_sync_late_arrival_done, dsa,
+	    dmu_sync_late_arrival_ready, NULL, dmu_sync_late_arrival_done, dsa,
 	    ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL | ZIO_FLAG_FASTWRITE, zb));
 
 	return (0);
@@ -1699,8 +1703,9 @@ dmu_sync(zio_t *pio, uint64_t txg, dmu_sync_cb_t *done, zgd_t *zgd)
 
 	zio_nowait(arc_write(pio, os->os_spa, txg,
 	    bp, dr->dt.dl.dr_data, DBUF_IS_L2CACHEABLE(db),
-	    DBUF_IS_L2COMPRESSIBLE(db), &zp, dmu_sync_ready, dmu_sync_done,
-	    dsa, ZIO_PRIORITY_SYNC_WRITE, ZIO_FLAG_CANFAIL | ZIO_FLAG_FASTWRITE, &zb));
+	    DBUF_IS_L2COMPRESSIBLE(db), &zp, dmu_sync_ready,
+	    NULL, dmu_sync_done, dsa, ZIO_PRIORITY_SYNC_WRITE,
+	    ZIO_FLAG_CANFAIL, &zb));
 
 	return (0);
 }