Fix log vdev removal issues

When we clear the log, we should clear all the fields, not only
zh_log.  Otherwise remaining ZIL_REPLAY_NEEDED will prevent the
vdev removal.  Handle it also from the other side, when zh_log
is already cleared, while zh_flags is not.

spa_vdev_remove_log() asserts that allocated space on removed log
device is zero.  While it should be so in perfect world, it might
be not if space leaked at any point.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Alexander Motin <alexander.motin@TrueNAS.com>
Closes #18277

module/zfs/vdev_removal.c

@@ -2151,7 +2151,6 @@ spa_vdev_remove_log(vdev_t *vd, uint64_t *txg)
 		ASSERT0P(vd->vdev_log_mg);
 		return (error);
 	}
-	ASSERT0(vd->vdev_stat.vs_alloc);
 
 	/*
 	 * The evacuation succeeded.  Remove any remaining MOS metadata

module/zfs/zil.c

@@ -1096,7 +1096,7 @@ zil_destroy(zilog_t *zilog, boolean_t keep_first)
 
 	zilog->zl_old_header = *zh;		/* debugging aid */
 
-	if (BP_IS_HOLE(&zh->zh_log))
+	if (BP_IS_HOLE(&zh->zh_log) && zh->zh_flags == 0)
 		return (B_FALSE);
 
 	tx = dmu_tx_create(zilog->zl_os);
@@ -1166,6 +1166,15 @@ zil_claim(dsl_pool_t *dp, dsl_dataset_t *ds, void *txarg)
 	zilog = dmu_objset_zil(os);
 	zh = zil_header_in_syncing_context(zilog);
 	ASSERT3U(tx->tx_txg, ==, spa_first_txg(zilog->zl_spa));
+
+	/*
+	 * If the log is empty, then there is nothing to do here.
+	 */
+	if (BP_IS_HOLE(&zh->zh_log)) {
+		dmu_objset_disown(os, B_FALSE, FTAG);
+		return (0);
+	}
+
 	first_txg = spa_min_claim_txg(zilog->zl_spa);
 
 	/*
@@ -1198,11 +1207,14 @@ zil_claim(dsl_pool_t *dp, dsl_dataset_t *ds, void *txarg)
 	if (spa_get_log_state(zilog->zl_spa) == SPA_LOG_CLEAR ||
 	    (zilog->zl_spa->spa_uberblock.ub_checkpoint_txg != 0 &&
 	    zh->zh_claim_txg == 0)) {
-		if (!BP_IS_HOLE(&zh->zh_log)) {
+		if (zilog->zl_spa->spa_uberblock.ub_checkpoint_txg != 0 &&
+		    BP_GET_BIRTH(&zh->zh_log) < first_txg) {
 			(void) zil_parse(zilog, zil_clear_log_block,
 			    zil_noop_log_record, tx, first_txg, B_FALSE);
+		} else {
+			zio_free(zilog->zl_spa, first_txg, &zh->zh_log);
 		}
-		BP_ZERO(&zh->zh_log);
+		memset(zh, 0, sizeof (zil_header_t));
 		if (os->os_encrypted)
 			os->os_next_write_raw[tx->tx_txg & TXG_MASK] = B_TRUE;
 		dsl_dataset_dirty(dmu_objset_ds(os), tx);
@@ -1224,7 +1236,7 @@ zil_claim(dsl_pool_t *dp, dsl_dataset_t *ds, void *txarg)
 	 * or destroy beyond the last block we successfully claimed.
 	 */
 	ASSERT3U(zh->zh_claim_txg, <=, first_txg);
-	if (zh->zh_claim_txg == 0 && !BP_IS_HOLE(&zh->zh_log)) {
+	if (zh->zh_claim_txg == 0) {
 		(void) zil_parse(zilog, zil_claim_log_block,
 		    zil_claim_log_record, tx, first_txg, B_FALSE);
 		zh->zh_claim_txg = first_txg;

tests/runfiles/common.run

@@ -947,7 +947,7 @@ tests = ['removal_all_vdev', 'removal_cancel', 'removal_check_space',
     'removal_with_write', 'removal_with_zdb', 'remove_expanded',
     'remove_mirror', 'remove_mirror_sanity', 'remove_raidz',
     'remove_indirect', 'remove_attach_mirror', 'removal_reservation',
-    'removal_with_hole']
+    'removal_with_hole', 'removal_with_missing_log']
 tags = ['functional', 'removal']
 
 [tests/functional/rename_dirs]

tests/zfs-tests/tests/Makefile.am

@@ -1947,6 +1947,7 @@ nobase_dist_datadir_zfs_tests_tests_SCRIPTS += \
 	functional/removal/removal_with_faulted.ksh \
 	functional/removal/removal_with_ganging.ksh \
 	functional/removal/removal_with_hole.ksh \
+	functional/removal/removal_with_missing_log.ksh \
 	functional/removal/removal_with_indirect.ksh \
 	functional/removal/removal_with_remove.ksh \
 	functional/removal/removal_with_scrub.ksh \

tests/zfs-tests/tests/functional/removal/removal_with_missing_log.ksh

@@ -0,0 +1,95 @@
+#!/bin/ksh -p
+# SPDX-License-Identifier: CDDL-1.0
+#
+# 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) 2026, TrueNAS.
+#
+
+. $STF_SUITE/include/libtest.shlib
+. $STF_SUITE/tests/functional/removal/removal.kshlib
+
+#
+# DESCRIPTION:
+#	Verify that a missing SLOG device can be removed even when
+#	ZIL blocks exist on it.
+#
+# STRATEGY:
+#	1. Create a pool with a SLOG device
+#	2. Freeze the pool and write data to ZIL
+#	3. Export the pool (ZIL blocks remain uncommitted)
+#	4. Import with -N to claim logs without replay
+#	5. Export and clear SLOG device labels to simulate failure
+#	6. Import with -m (missing devices allowed)
+#	7. Remove the missing SLOG vdev
+#	8. Verify pool is healthy and space accounting is correct
+#
+
+verify_runnable "global"
+
+log_assert "Removal of missing SLOG with ZIL blocks succeeds"
+
+function cleanup
+{
+	poolexists $TESTPOOL && destroy_pool $TESTPOOL
+}
+
+log_onexit cleanup
+
+VDEV1="$(echo $DISKS | cut -d' ' -f1)"
+VDEV2="$(echo $DISKS | cut -d' ' -f2)"
+
+# Create pool with SLOG and dataset
+log_must zpool create $TESTPOOL $VDEV1 log $VDEV2
+log_must zfs create $TESTPOOL/$TESTFS
+
+# Create initial ZIL header (required before freezing)
+log_must dd if=/dev/zero of=/$TESTPOOL/$TESTFS/init \
+    conv=fdatasync,fsync bs=1 count=1
+
+# Freeze pool and write data to ZIL
+log_must zpool freeze $TESTPOOL
+log_must dd if=/dev/urandom of=/$TESTPOOL/$TESTFS/file1 \
+    oflag=sync bs=128k count=128
+
+# Export with uncommitted ZIL transactions
+log_must zpool export $TESTPOOL
+
+# Import with -N to claim logs without mounting/replaying
+log_must zpool import -N $TESTPOOL
+log_must zpool export $TESTPOOL
+
+# Clear SLOG labels to simulate device failure
+log_must zpool labelclear -f $VDEV2
+
+# Import with missing SLOG allowed
+log_must zpool import -m $TESTPOOL
+log_must eval "zpool status $TESTPOOL | grep UNAVAIL"
+
+# Remove the missing SLOG - should succeed
+log_must zpool remove $TESTPOOL $VDEV2
+log_must zpool wait -t remove $TESTPOOL
+sync_pool $TESTPOOL
+log_mustnot eval "zpool status -v $TESTPOOL | grep $VDEV2"
+
+# Verify pool health
+log_must zpool scrub -w $TESTPOOL
+log_must check_pool_status $TESTPOOL "errors" "No known data errors"
+
+# Verify space accounting is correct
+log_must zdb -c $TESTPOOL
+
+log_pass "Removal of missing SLOG with ZIL blocks succeeded"