zfsonlinux/zfs-patches/0047-Change-checksum-IO-delay-ratelimit-values.patch

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Tony Hutter <hutter2@llnl.gov>
Date: Sun, 4 Mar 2018 17:34:51 -0800
Subject: [PATCH] Change checksum & IO delay ratelimit values
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Change checksum & IO delay ratelimit thresholds from 5/sec to 20/sec.
This allows zed to actually trigger if a bunch of these events arrive in
a short period of time (zed has a threshold of 10 events in 10 sec).
Previously, if you had, say, 100 checksum errors in 1 sec, it would get
ratelimited to 5/sec which wouldn't trigger zed to fault the drive.

Also, convert the checksum and IO delay thresholds to module params for
easy testing.

Reviewed-by: loli10K <ezomori.nozomu@gmail.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Giuseppe Di Natale <dinatale2@llnl.gov>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #7252
(cherry picked from commit 6dc40e2ada2d0d008bd314ff3525f2b0acc2bb01)
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
---
 include/sys/vdev_impl.h          |  2 --
 include/sys/zfs_ratelimit.h      | 12 +++++++++---
 module/zcommon/zfs_comutil.c     |  4 ++--
 module/zfs/vdev.c                | 23 +++++++++++++++++++++--
 man/man5/zfs-module-parameters.5 | 39 +++++++++++++++++++++++++++++++++++++++
 5 files changed, 71 insertions(+), 9 deletions(-)

diff --git a/include/sys/vdev_impl.h b/include/sys/vdev_impl.h
index 13c495822..4f9f1a903 100644
--- a/include/sys/vdev_impl.h
+++ b/include/sys/vdev_impl.h
@@ -255,8 +255,6 @@ struct vdev {
 	 * We rate limit ZIO delay and ZIO checksum events, since they
 	 * can flood ZED with tons of events when a drive is acting up.
 	 */
-#define	DELAYS_PER_SECOND 5
-#define	CHECKSUMS_PER_SECOND 5
 	zfs_ratelimit_t vdev_delay_rl;
 	zfs_ratelimit_t vdev_checksum_rl;
 };
diff --git a/include/sys/zfs_ratelimit.h b/include/sys/zfs_ratelimit.h
index f36e07841..012825fad 100644
--- a/include/sys/zfs_ratelimit.h
+++ b/include/sys/zfs_ratelimit.h
@@ -25,13 +25,19 @@
 typedef struct {
 	hrtime_t start;
 	unsigned int count;
-	unsigned int burst;		/* Number to allow per interval */
-	unsigned int interval;		/* Interval length in seconds */
+
+	/*
+	 * Pointer to number of events per interval.  We do this to
+	 * allow the burst to be a (changeable) module parameter.
+	 */
+	unsigned int *burst;
+
+	unsigned int interval;	/* Interval length in seconds */
 	kmutex_t lock;
 } zfs_ratelimit_t;
 
 int zfs_ratelimit(zfs_ratelimit_t *rl);
-void zfs_ratelimit_init(zfs_ratelimit_t *rl, unsigned int burst,
+void zfs_ratelimit_init(zfs_ratelimit_t *rl, unsigned int *burst,
     unsigned int interval);
 void zfs_ratelimit_fini(zfs_ratelimit_t *rl);
 
diff --git a/module/zcommon/zfs_comutil.c b/module/zcommon/zfs_comutil.c
index 52cb7e365..44cdc8523 100644
--- a/module/zcommon/zfs_comutil.c
+++ b/module/zcommon/zfs_comutil.c
@@ -215,7 +215,7 @@ const char *zfs_history_event_names[ZFS_NUM_LEGACY_HISTORY_EVENTS] = {
  * interval:	Interval time in seconds
  */
 void
-zfs_ratelimit_init(zfs_ratelimit_t *rl, unsigned int burst,
+zfs_ratelimit_init(zfs_ratelimit_t *rl, unsigned int *burst,
     unsigned int interval)
 {
 	rl->count = 0;
@@ -270,7 +270,7 @@ zfs_ratelimit(zfs_ratelimit_t *rl)
 		rl->start = now;
 		rl->count = 0;
 	} else {
-		if (rl->count >= rl->burst) {
+		if (rl->count >= *rl->burst) {
 			rc = 0;	/* We're ratelimiting */
 		}
 	}
diff --git a/module/zfs/vdev.c b/module/zfs/vdev.c
index df07d893d..0786fbb83 100644
--- a/module/zfs/vdev.c
+++ b/module/zfs/vdev.c
@@ -56,6 +56,16 @@
  */
 int metaslabs_per_vdev = 200;
 
+/*
+ * Rate limit delay events to this many IO delays per second.
+ */
+unsigned int zfs_delays_per_second = 20;
+
+/*
+ * Rate limit checksum events after this many checksum errors per second.
+ */
+unsigned int zfs_checksums_per_second = 20;
+
 /*
  * Virtual device management.
  */
@@ -357,8 +367,8 @@ vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid, vdev_ops_t *ops)
 	 * and checksum events so that we don't overwhelm ZED with thousands
 	 * of events when a disk is acting up.
 	 */
-	zfs_ratelimit_init(&vd->vdev_delay_rl, DELAYS_PER_SECOND, 1);
-	zfs_ratelimit_init(&vd->vdev_checksum_rl, CHECKSUMS_PER_SECOND, 1);
+	zfs_ratelimit_init(&vd->vdev_delay_rl, &zfs_delays_per_second, 1);
+	zfs_ratelimit_init(&vd->vdev_checksum_rl, &zfs_checksums_per_second, 1);
 
 	list_link_init(&vd->vdev_config_dirty_node);
 	list_link_init(&vd->vdev_state_dirty_node);
@@ -3776,5 +3786,14 @@ module_param(metaslabs_per_vdev, int, 0644);
 MODULE_PARM_DESC(metaslabs_per_vdev,
 	"Divide added vdev into approximately (but no more than) this number "
 	"of metaslabs");
+
+module_param(zfs_delays_per_second, uint, 0644);
+MODULE_PARM_DESC(zfs_delays_per_second, "Rate limit delay events to this many "
+	"IO delays per second");
+
+module_param(zfs_checksums_per_second, uint, 0644);
+	MODULE_PARM_DESC(zfs_checksums_per_second, "Rate limit checksum events "
+	"to this many checksum errors per second (do not set below zed"
+	"threshold).");
 /* END CSTYLED */
 #endif
diff --git a/man/man5/zfs-module-parameters.5 b/man/man5/zfs-module-parameters.5
index d4daffde6..8d5ac2576 100644
--- a/man/man5/zfs-module-parameters.5
+++ b/man/man5/zfs-module-parameters.5
@@ -739,6 +739,34 @@ Disable pool import at module load by ignoring the cache file (typically \fB/etc
 Use \fB1\fR for yes (default) and \fB0\fR for no.
 .RE
 
+.sp
+.ne 2
+.na
+\fBzfs_checksums_per_second\fR (int)
+.ad
+.RS 12n
+Rate limit checksum events to this many per second.  Note that this should
+not be set below the zed thresholds (currently 10 checksums over 10 sec)
+or else zed may not trigger any action.
+.sp
+Default value: 20
+.RE
+
+.sp
+.ne 2
+.na
+\fBzfs_commit_timeout_pct\fR (int)
+.ad
+.RS 12n
+This controls the amount of time that a ZIL block (lwb) will remain "open"
+when it isn't "full", and it has a thread waiting for it to be committed to
+stable storage.  The timeout is scaled based on a percentage of the last lwb
+latency to avoid significantly impacting the latency of each individual
+transaction record (itx).
+.sp
+Default value: \fB5\fR%.
+.RE
+
 .sp
 .ne 2
 .na
@@ -866,6 +894,17 @@ Note: \fBzfs_delay_scale\fR * \fBzfs_dirty_data_max\fR must be < 2^64.
 Default value: \fB500,000\fR.
 .RE
 
+.sp
+.ne 2
+.na
+\fBzfs_delays_per_second\fR (int)
+.ad
+.RS 12n
+Rate limit IO delay events to this many per second.
+.sp
+Default value: 20
+.RE
+
 .sp
 .ne 2
 .na
-- 
2.14.2