ZED should handle spares using configured ashift

If the zpool 'ashift' property is set then ZED should use its value
when kicking in a hotspare; with this change 512e disks can be used
as spares for VDEVs that were created with ashift=9, even if ZFS
natively detects them as 4K block devices.

Signed-off-by: loli10K <ezomori.nozomu@gmail.com>
Requires-builders: style test

TEST

@@ -5,25 +5,25 @@
 #TEST_PREPARE_SHARES="yes"
 
 ### SPLAT
-#TEST_SPLAT_SKIP="yes"
+TEST_SPLAT_SKIP="yes"
 #TEST_SPLAT_OPTIONS="-acvx"
 
 ### ztest
-#TEST_ZTEST_SKIP="yes"
+TEST_ZTEST_SKIP="yes"
 #TEST_ZTEST_TIMEOUT=1800
 #TEST_ZTEST_DIR="/var/tmp/"
 #TEST_ZTEST_OPTIONS="-V"
 #TEST_ZTEST_CORE_DIR="/mnt/zloop"
 
 ### zimport
-#TEST_ZIMPORT_SKIP="yes"
+TEST_ZIMPORT_SKIP="yes"
 #TEST_ZIMPORT_DIR="/var/tmp/zimport"
 #TEST_ZIMPORT_VERSIONS="master installed"
 #TEST_ZIMPORT_POOLS="zol-0.6.1 zol-0.6.2 master installed"
 #TEST_ZIMPORT_OPTIONS="-c"
 
 ### xfstests
-#TEST_XFSTESTS_SKIP="yes"
+TEST_XFSTESTS_SKIP="yes"
 #TEST_XFSTESTS_URL="https://github.com/behlendorf/xfstests/archive/"
 #TEST_XFSTESTS_VER="zfs.tar.gz"
 #TEST_XFSTESTS_POOL="tank"
@@ -39,10 +39,11 @@
 #TEST_ZFSTESTS_ITERS="1"
 #TEST_ZFSTESTS_OPTIONS="-vx"
 #TEST_ZFSTESTS_RUNFILE="linux.run"
+TEST_ZFSTESTS_RUNFILE="issue-2562.run"
 #TEST_ZFSTESTS_TAGS="functional"
 
 ### zfsstress
-#TEST_ZFSSTRESS_SKIP="yes"
+TEST_ZFSSTRESS_SKIP="yes"
 #TEST_ZFSSTRESS_URL="https://github.com/nedbass/zfsstress/archive/"
 #TEST_ZFSSTRESS_VER="master.tar.gz"
 #TEST_ZFSSTRESS_RUNTIME=300

cmd/zed/agents/zfs_retire.c

@@ -176,6 +176,8 @@ replace_with_spare(fmd_hdl_t *hdl, zpool_handle_t *zhp, nvlist_t *vdev)
 	nvlist_t **spares;
 	uint_t s, nspares;
 	char *dev_name;
+	zprop_source_t source;
+	int ashift;
 
 	config = zpool_get_config(zhp, NULL);
 	if (nvlist_lookup_nvlist(config, ZPOOL_CONFIG_VDEV_TREE,
@@ -189,6 +191,11 @@ replace_with_spare(fmd_hdl_t *hdl, zpool_handle_t *zhp, nvlist_t *vdev)
 	    &spares, &nspares) != 0)
 		return;
 
+	/*
+	 * lookup "ashift" pool property, we may need it for the replacement
+	 */
+	ashift = zpool_get_prop_int(zhp, ZPOOL_PROP_ASHIFT, &source);
+
 	replacement = fmd_nvl_alloc(hdl, FMD_SLEEP);
 
 	(void) nvlist_add_string(replacement, ZPOOL_CONFIG_TYPE,
@@ -207,6 +214,11 @@ replace_with_spare(fmd_hdl_t *hdl, zpool_handle_t *zhp, nvlist_t *vdev)
 		    &spare_name) != 0)
 			continue;
 
+		/* if set, add the "ashift" pool property to the spare nvlist */
+		if (source != ZPROP_SRC_DEFAULT)
+			(void) nvlist_add_uint64(spares[s],
+			    ZPOOL_CONFIG_ASHIFT, ashift);
+
 		(void) nvlist_add_nvlist_array(replacement,
 		    ZPOOL_CONFIG_CHILDREN, &spares[s], 1);
 

tests/runfiles/issue-2562.run

@@ -0,0 +1,37 @@
+#
+# This file and its contents are supplied under the terms of the
+# Common Development and Distribution License ("CDDL"), version 1.0.
+# You may only use this file in accordance with the terms of version
+# 1.0 of the CDDL.
+#
+# A full copy of the text of the CDDL should have accompanied this
+# source.  A copy of the CDDL is also available via the Internet at
+# http://www.illumos.org/license/CDDL.
+#
+
+[DEFAULT]
+pre = setup
+quiet = False
+pre_user = root
+user = root
+timeout = 600
+post_user = root
+post = cleanup
+outputdir = /var/tmp/test_results
+
+[tests/functional/fault]
+tests = ['auto_online_001_pos', 'auto_replace_001_pos', 'auto_spare_001_pos',
+    'auto_spare_002_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos',
+    'auto_spare_003_pos', 'auto_spare_003_pos', 'auto_spare_003_pos']
+tags = ['functional', 'fault']
+

tests/runfiles/linux.run

@@ -464,7 +464,7 @@ tags = ['functional', 'exec']
 
 [tests/functional/fault]
 tests = ['auto_online_001_pos', 'auto_replace_001_pos', 'auto_spare_001_pos',
-    'auto_spare_002_pos.ksh']
+    'auto_spare_002_pos', 'auto_spare_003_pos']
 tags = ['functional', 'fault']
 
 [tests/functional/features/async_destroy]

tests/zfs-tests/tests/functional/fault/Makefile.am

@@ -6,4 +6,5 @@ dist_pkgdata_SCRIPTS = \
 	auto_online_001_pos.ksh \
 	auto_replace_001_pos.ksh \
 	auto_spare_001_pos.ksh \
-	auto_spare_002_pos.ksh
+	auto_spare_002_pos.ksh \
+	auto_spare_003_pos.ksh

tests/zfs-tests/tests/functional/fault/auto_spare_003_pos.ksh

@@ -0,0 +1,99 @@
+#!/bin/ksh -p
+
+#
+# CDDL HEADER START
+#
+# This file and its contents are supplied under the terms of the
+# Common Development and Distribution License ("CDDL"), version 1.0.
+# You may only use this file in accordance with the terms of version
+# 1.0 of the CDDL.
+#
+# A full copy of the text of the CDDL should have accompanied this
+# source.  A copy of the CDDL is also available via the Internet at
+# http://www.illumos.org/license/CDDL.
+#
+# CDDL HEADER END
+#
+
+#
+# Copyright (c) 2017 by Intel Corporation. All rights reserved.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/include/math.shlib
+. $STF_SUITE/tests/functional/fault/fault.cfg
+
+#
+# DESCRIPTION:
+# Testing Fault Management Agent ZED Logic - Automated Auto-Spare Test when
+# drive is faulted and a custom ashift value needs to be provided to replace it.
+#
+# STRATEGY:
+# 1. Create a pool from 512b devices and set "ashift" pool property accordingly
+# 2. Add one 512e spare device (4Kn would generate IO errors on replace)
+# 3. Inject IO errors with a zinject error handler
+# 4. Start a scrub
+# 5. Verify the ZED kicks in the hot spare and expected pool/device status
+# 6. Clear the fault
+# 7. Verify the hot spare is available and expected pool/device status
+#
+
+verify_runnable "both"
+
+function cleanup
+{
+	log_must zinject -c all
+	destroy_pool $TESTPOOL
+	rm -f $SAFE_DEVICE $FAIL_DEVICE
+}
+
+log_assert "ZED should replace a device using the configured ashift property"
+log_onexit cleanup
+
+SAFE_DEVICE="$TEST_BASE_DIR/safe-dev"
+FAIL_DEVICE="$TEST_BASE_DIR/fail-dev"
+
+# 1. Create a pool from 512b devices and set "ashift" pool property accordingly
+for vdev in $SAFE_DEVICE $FAIL_DEVICE; do
+	truncate -s $SPA_MINDEVSIZE $vdev
+done
+log_must zpool create -f $TESTPOOL mirror $SAFE_DEVICE $FAIL_DEVICE
+# NOTE: file VDEVs should be added as 512b devices, verify this "just in case"
+#for vdev in $SAFE_DEVICE $FAIL_DEVICE; do
+# XXX: math.shlib	verify_eq "9" "$(zdb -e -l $vdev | awk '/ashift: /{print $2}')" "ashift"
+#done
+log_must zpool set ashift=9 $TESTPOOL
+
+# 2. Add one 512e spare device (4Kn would generate IO errors on replace)
+# NOTE: must be larger than the existing 512b devices, add 32m of fudge
+# XXX: update load_scsi_debug (blkdev.shlib) to handle sector_size/physblk_exp params
+modprobe -r scsi_debug
+log_must modprobe scsi_debug dev_size_mb=$(($SPA_MINDEVSIZE/1024/1024+32)) add_host=1 num_tgts=1 max_luns=1 sector_size=512 physblk_exp=3
+block_device_wait
+SPARE_DEVICE=$(get_debug_device)
+log_must zpool add $TESTPOOL spare $SPARE_DEVICE
+
+# 3. Inject IO errors with a zinject error handler
+log_must zinject -d $FAIL_DEVICE -e io -T all -f 100 $TESTPOOL
+
+# 4. Start a scrub
+log_must zpool scrub $TESTPOOL
+
+# 5. Verify the ZED kicks in a hot spare and expected pool/device status
+log_note "Wait for ZED to auto-spare"
+log_must wait_vdev_state $TESTPOOL $FAIL_DEVICE "FAULTED" 60
+log_must wait_vdev_state $TESTPOOL $SPARE_DEVICE "ONLINE" 60
+log_must wait_hotspare_state $TESTPOOL $SPARE_DEVICE "INUSE"
+log_must check_state $TESTPOOL "" "DEGRADED"
+
+# 6. Clear the fault
+log_must zinject -c all
+log_must zpool clear $TESTPOOL $FAIL_DEVICE
+
+# 7. Verify the hot spare is available and expected pool/device status
+log_must wait_vdev_state $TESTPOOL $FAIL_DEVICE "ONLINE" 60
+log_must wait_hotspare_state $TESTPOOL $SPARE_DEVICE "AVAIL"
+log_must is_pool_resilvered $TESTPOOL
+log_must check_state $TESTPOOL "" "ONLINE"
+
+log_pass "ZED successfully replaces a device using the configured ashift property"