Add scrub after resilver zed script

* Add a zed script to kick off a scrub after a resilver.  The script is
disabled by default.

* Add a optional $PATH (-P) option to zed to allow it to use a custom
$PATH for its zedlets.  This is needed when you're running zed under
the ZTS in a local workspace.

* Update test scripts to not copy in all-debug.sh and all-syslog.sh by
default.  They can be optionally copied in as part of zed_setup().
These scripts slow down zed considerably under heavy events loads and
can cause events to be dropped or their delivery delayed. This was
causing some sporadic failures in the 'fault' tests.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Richard Laager <rlaager@wiktel.com>
Signed-off-by: Tony Hutter <hutter2@llnl.gov>
Closes #4662 
Closes #7086
This commit is contained in:
Tony Hutter
2018-02-23 11:38:05 -08:00
committed by Brian Behlendorf
parent e9a7729008
commit bf95a000c4
16 changed files with 187 additions and 21 deletions
+1
View File
@@ -86,6 +86,7 @@ export SYSTEM_FILES='arp
pgrep
ping
pkill
printenv
printf
ps
pwd
+43 -7
View File
@@ -3053,9 +3053,32 @@ function wait_replacing #pool
done
}
#
# Wait for a pool to be scrubbed
#
# $1 pool name
# $2 number of seconds to wait (optional)
#
# Returns true when pool has been scrubbed, or false if there's a timeout or if
# no scrub was done.
#
function wait_scrubbed
{
typeset pool=${1:-$TESTPOOL}
typeset iter=${2:-10}
for i in {1..$iter} ; do
if is_pool_scrubbed $pool ; then
return 0
fi
sleep 1
done
return 1
}
#
# Setup custom environment for the ZED.
#
# $@ Optional list of zedlets to run under zed.
function zed_setup
{
if ! is_linux; then
@@ -3073,6 +3096,7 @@ function zed_setup
if [[ -e $VDEVID_CONF_ETC ]]; then
log_fail "Must not have $VDEVID_CONF_ETC file present on system"
fi
EXTRA_ZEDLETS=$@
# Create a symlink for /etc/zfs/vdev_id.conf file.
log_must ln -s $VDEVID_CONF $VDEVID_CONF_ETC
@@ -3082,32 +3106,44 @@ function zed_setup
log_must cp ${ZEDLET_ETC_DIR}/zed.rc $ZEDLET_DIR
log_must cp ${ZEDLET_ETC_DIR}/zed-functions.sh $ZEDLET_DIR
# Scripts must only be user writable.
if [[ -n "$EXTRA_ZEDLETS" ]] ; then
saved_umask=$(umask)
log_must umask 0022
for i in $EXTRA_ZEDLETS ; do
log_must cp ${ZEDLET_LIBEXEC_DIR}/$i $ZEDLET_DIR
done
log_must umask $saved_umask
fi
# Customize the zed.rc file to enable the full debug log.
log_must sed -i '/\#ZED_DEBUG_LOG=.*/d' $ZEDLET_DIR/zed.rc
echo "ZED_DEBUG_LOG=$ZED_DEBUG_LOG" >>$ZEDLET_DIR/zed.rc
# Scripts must only be user writable.
saved_umask=$(umask)
log_must umask 0022
log_must cp ${ZEDLET_LIBEXEC_DIR}/all-syslog.sh $ZEDLET_DIR
log_must cp ${ZEDLET_LIBEXEC_DIR}/all-debug.sh $ZEDLET_DIR
log_must umask $saved_umask
}
#
# Cleanup custom ZED environment.
#
# $@ Optional list of zedlets to remove from our test zed.d directory.
function zed_cleanup
{
if ! is_linux; then
return
fi
EXTRA_ZEDLETS=$@
log_must rm -f ${ZEDLET_DIR}/zed.rc
log_must rm -f ${ZEDLET_DIR}/zed-functions.sh
log_must rm -f ${ZEDLET_DIR}/all-syslog.sh
log_must rm -f ${ZEDLET_DIR}/all-debug.sh
log_must rm -f ${ZEDLET_DIR}/state
if [[ -n "$EXTRA_ZEDLETS" ]] ; then
for i in $EXTRA_ZEDLETS ; do
log_must rm -f ${ZEDLET_DIR}/$i
done
fi
log_must rm -f $ZED_LOG
log_must rm -f $ZED_DEBUG_LOG
log_must rm -f $VDEVID_CONF_ETC
@@ -3139,7 +3175,7 @@ function zed_start
# run ZED in the background and redirect foreground logging
# output to $ZED_LOG.
log_must truncate -s 0 $ZED_DEBUG_LOG
log_must eval "zed -vF -d $ZEDLET_DIR -p $ZEDLET_DIR/zed.pid" \
log_must eval "zed -vF -d $ZEDLET_DIR -p $ZEDLET_DIR/zed.pid -P $PATH" \
"-s $ZEDLET_DIR/state 2>$ZED_LOG &"
return 0
@@ -26,6 +26,6 @@
. $STF_SUITE/include/libtest.shlib
zed_cleanup
zed_cleanup all-debug.sh all-syslog.sh
default_cleanup
@@ -28,6 +28,6 @@
DISK=${DISKS%% *}
zed_setup
zed_setup all-debug.sh all-syslog.sh
default_setup $DISK
@@ -8,4 +8,5 @@ dist_pkgdata_SCRIPTS = \
auto_spare_001_pos.ksh \
auto_spare_002_pos.ksh \
auto_spare_ashift.ksh \
auto_spare_multiple.ksh
auto_spare_multiple.ksh \
scrub_after_resilver.ksh
@@ -31,6 +31,6 @@ verify_runnable "global"
cleanup_devices $DISKS
zed_stop
zed_cleanup
zed_cleanup resilver_finish-start-scrub.sh
log_pass
@@ -0,0 +1,65 @@
#!/bin/ksh -p
#
# 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.
#
#
# Copyright (c) 2018 by Lawrence Livermore National Security, LLC.
# All rights reserved.
#
. $STF_SUITE/include/libtest.shlib
. $STF_SUITE/tests/functional/fault/fault.cfg
#
# DESCRIPTION:
# Test the scrub after resilver zedlet
#
# STRATEGY:
# 1. Create a mirrored pool
# 2. Fault a disk
# 3. Replace the disk, starting a resilver
# 4. Verify that a scrub happens after the resilver finishes
#
log_assert "Testing the scrub after resilver zedlet"
# Backup our zed.rc
zedrc_backup="$(mktemp)"
log_must cp $ZEDLET_DIR/zed.rc $zedrc_backup
# Enable ZED_SCRUB_AFTER_RESILVER
eval "sed -i 's/\#ZED_SCRUB_AFTER_RESILVER/ZED_SCRUB_AFTER_RESILVER/g' $ZEDLET_DIR/zed.rc"
function cleanup
{
# Restore our zed.rc
log_must mv $zedrc_backup $ZEDLET_DIR/zed.rc
default_cleanup_noexit
}
log_onexit cleanup
verify_disk_count "$DISKS" 3
default_mirror_setup_noexit $DISK1 $DISK2
log_must zpool offline -f $TESTPOOL $DISK1
# Write to our degraded pool so we have some data to resilver
log_must mkfile 16M $TESTDIR/file1
# Replace the failed disks, forcing a resilver
log_must zpool replace $TESTPOOL $DISK1 $DISK3
# Wait for the resilver to finish, and then the subsequent scrub to finish.
# Waiting for the scrub has the effect of waiting for both. Timeout after 10
# seconds if nothing is happening.
log_must wait_scrubbed $TESTPOOL 10
log_pass "Successfully ran the scrub after resilver zedlet"
@@ -28,7 +28,7 @@
verify_runnable "global"
zed_setup
zed_setup resilver_finish-start-scrub.sh
zed_start
log_pass