OpenZFS 9617 - too-frequent TXG sync causes excessive write inflation

Porting notes:
* Renamed zfs_dirty_data_sync_pct to zfs_dirty_data_sync_percent and
  changed the type to be consistent with the other dirty module params.
* Updated zfs-module-parameters.5 accordingly.

Authored by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Serapheim Dimitropoulos <serapheim.dimitro@delphix.com>
Reviewed by: Brad Lewis <brad.lewis@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Andrew Stormont <andyjstormont@gmail.com>
Reviewed-by: George Melikov <mail@gmelikov.ru>
Approved by: Robert Mustacchi <rm@joyent.com>
Ported-by: Brian Behlendorf <behlendorf1@llnl.gov>
OpenZFS-issue: https://illumos.org/issues/9617
OpenZFS-commit: https://github.com/openzfs/openzfs/commit/7928f4ba
Closes #7976

diff --git a/include/sys/dsl_pool.h b/include/sys/dsl_pool.h
index 01870e8670466190692cf4810dead71350d04a06..56317cf733f91cc18f02e9ed16941aa616ca10b2 100644 (file)
--- a/include/sys/dsl_pool.h
+++ b/include/sys/dsl_pool.h
@@ -57,7 +57,7 @@ struct dsl_crypto_params;
 
 extern unsigned long zfs_dirty_data_max;
 extern unsigned long zfs_dirty_data_max_max;
-extern unsigned long zfs_dirty_data_sync;
+extern int zfs_dirty_data_sync_percent;
 extern int zfs_dirty_data_max_percent;
 extern int zfs_dirty_data_max_max_percent;
 extern int zfs_delay_min_dirty_percent;

diff --git a/man/man5/zfs-module-parameters.5 b/man/man5/zfs-module-parameters.5
index b02d6aae736a22e0093ca53b453afb2d5400cd36..d7b53d161789ec7574bfec7931677239c4b89060 100644 (file)
--- a/man/man5/zfs-module-parameters.5
+++ b/man/man5/zfs-module-parameters.5
@@ -1225,12 +1225,14 @@ Default value: \fB10\fR%, subject to \fBzfs_dirty_data_max_max\fR.
 .sp
 .ne 2
 .na
-\fBzfs_dirty_data_sync\fR (int)
+\fBzfs_dirty_data_sync_percent\fR (int)
 .ad
 .RS 12n
-Start syncing out a transaction group if there is at least this much dirty data.
+Start syncing out a transaction group if there's at least this much dirty data
+as a percentage of \fBzfs_dirty_data_max\fR.  This should be less than
+\fBzfs_vdev_async_write_active_min_dirty_percent\fR.
 .sp
-Default value: \fB67,108,864\fR.
+Default value: \fB20\fR% of \fBzfs_dirty_data_max\fR.
 .RE
 
 .sp

diff --git a/module/zfs/dsl_pool.c b/module/zfs/dsl_pool.c
index bd205571058253513f8b8ea42d2ce51e8835e70e..b940382d1f912274f0977d2901cf21a3af96ca2b 100644 (file)
--- a/module/zfs/dsl_pool.c
+++ b/module/zfs/dsl_pool.c
@@ -106,9 +106,11 @@ int zfs_dirty_data_max_percent = 10;
 int zfs_dirty_data_max_max_percent = 25;
 
 /*
- * If there is at least this much dirty data, push out a txg.
+ * If there's at least this much dirty data (as a percentage of
+ * zfs_dirty_data_max), push out a txg.  This should be less than
+ * zfs_vdev_async_write_active_min_dirty_percent.
  */
-unsigned long zfs_dirty_data_sync = 64 * 1024 * 1024;
+int zfs_dirty_data_sync_percent = 20;
 
 /*
  * Once there is this amount of dirty data, the dmu_tx_delay() will kick in
@@ -879,10 +881,12 @@ dsl_pool_need_dirty_delay(dsl_pool_t *dp)
 {
        uint64_t delay_min_bytes =
            zfs_dirty_data_max * zfs_delay_min_dirty_percent / 100;
+       uint64_t dirty_min_bytes =
+           zfs_dirty_data_max * zfs_dirty_data_sync_percent / 100;
        boolean_t rv;
 
        mutex_enter(&dp->dp_lock);
-       if (dp->dp_dirty_total > zfs_dirty_data_sync)
+       if (dp->dp_dirty_total > dirty_min_bytes)
                txg_kick(dp);
        rv = (dp->dp_dirty_total > delay_min_bytes);
        mutex_exit(&dp->dp_lock);
@@ -1345,7 +1349,7 @@ module_param(zfs_dirty_data_max_max, ulong, 0444);
 MODULE_PARM_DESC(zfs_dirty_data_max_max,
        "zfs_dirty_data_max upper bound in bytes");
 
-module_param(zfs_dirty_data_sync, ulong, 0644);
+module_param(zfs_dirty_data_sync_percent, int, 0644);
 MODULE_PARM_DESC(zfs_dirty_data_sync, "sync txg when this much dirty data");
 
 module_param(zfs_delay_scale, ulong, 0644);

diff --git a/module/zfs/txg.c b/module/zfs/txg.c
index cfc1d0a3514a015394457102402e4628c33ef577..aad1a8e572ea7c8ab8c48f2dcfeccd83a06a785b 100644 (file)
--- a/module/zfs/txg.c
+++ b/module/zfs/txg.c
@@ -517,7 +517,8 @@ txg_sync_thread(void *arg)
                clock_t timeout = zfs_txg_timeout * hz;
                clock_t timer;
                uint64_t txg;
-               txg_stat_t *ts;
+               uint64_t dirty_min_bytes =
+                   zfs_dirty_data_max * zfs_dirty_data_sync_percent / 100;
 
                /*
                 * We sync when we're scanning, there's someone waiting
@@ -529,7 +530,7 @@ txg_sync_thread(void *arg)
                    !tx->tx_exiting && timer > 0 &&
                    tx->tx_synced_txg >= tx->tx_sync_txg_waiting &&
                    !txg_has_quiesced_to_sync(dp) &&
-                   dp->dp_dirty_total < zfs_dirty_data_sync) {
+                   dp->dp_dirty_total < dirty_min_bytes) {
                        dprintf("waiting; tx_synced=%llu waiting=%llu dp=%p\n",
                            tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp);
                        txg_thread_wait(tx, &cpr, &tx->tx_sync_more_cv, timer);
@@ -561,7 +562,7 @@ txg_sync_thread(void *arg)
                tx->tx_quiesced_txg = 0;
                tx->tx_syncing_txg = txg;
                DTRACE_PROBE2(txg__syncing, dsl_pool_t *, dp, uint64_t, txg);
-               ts = spa_txg_history_init_io(spa, txg, dp);
+               txg_stat_t *ts = spa_txg_history_init_io(spa, txg, dp);
                cv_broadcast(&tx->tx_quiesce_more_cv);
 
                dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",