diff options
Diffstat (limited to 'qa/standalone/misc')
| -rwxr-xr-x | qa/standalone/misc/mclock-config.sh | 467 | ||||
| -rwxr-xr-x | qa/standalone/misc/network-ping.sh | 169 | ||||
| -rwxr-xr-x | qa/standalone/misc/ok-to-stop.sh | 296 | ||||
| -rwxr-xr-x | qa/standalone/misc/rados-striper.sh | 101 | ||||
| -rwxr-xr-x | qa/standalone/misc/test-ceph-helpers.sh | 21 | ||||
| -rwxr-xr-x | qa/standalone/misc/test-snaptrim-stats.sh | 188 | ||||
| -rwxr-xr-x | qa/standalone/misc/ver-health.sh | 231 |
7 files changed, 1473 insertions, 0 deletions
diff --git a/qa/standalone/misc/mclock-config.sh b/qa/standalone/misc/mclock-config.sh new file mode 100755 index 000000000..59f002584 --- /dev/null +++ b/qa/standalone/misc/mclock-config.sh @@ -0,0 +1,467 @@ +#!/usr/bin/env bash +# +# Copyright (C) 2022 Red Hat <contact@redhat.com> +# +# Author: Sridhar Seshasayee <sseshasa@redhat.com> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU Library Public License as published by +# the Free Software Foundation; either version 2, or (at your option) +# any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Library Public License for more details. +# + +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON="127.0.0.1:7124" # git grep '\<7124\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + CEPH_ARGS+="--mon-host=$CEPH_MON " + CEPH_ARGS+="--debug-mclock 20 " + + local funcs=${@:-$(set | sed -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} + for func in $funcs ; do + setup $dir || return 1 + $func $dir || return 1 + teardown $dir || return 1 + done +} + +function TEST_profile_builtin_to_custom() { + local dir=$1 + + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + + # Verify the default mclock profile on the OSD + local mclock_profile=$(ceph config get osd.0 osd_mclock_profile) + test "$mclock_profile" = "balanced" || return 1 + + # Verify the running mClock profile + mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile |\ + jq .osd_mclock_profile) + mclock_profile=$(eval echo $mclock_profile) + test "$mclock_profile" = "high_recovery_ops" || return 1 + + # Change the mclock profile to 'custom' + ceph tell osd.0 config set osd_mclock_profile custom || return 1 + + # Verify that the mclock profile is set to 'custom' on the OSDs + mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_profile | jq .osd_mclock_profile) + mclock_profile=$(eval echo $mclock_profile) + test "$mclock_profile" = "custom" || return 1 + + # Change a mclock config param and confirm the change + local client_res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + echo "client_res = $client_res" + local client_res_new=$(echo "$client_res + 0.1" | bc -l) + echo "client_res_new = $client_res_new" + ceph config set osd.0 osd_mclock_scheduler_client_res \ + $client_res_new || return 1 + + # Check value in config monitor db + local res=$(ceph config get osd.0 \ + osd_mclock_scheduler_client_res) || return 1 + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + # Check value in the in-memory 'values' map + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + + teardown $dir || return 1 +} + +function TEST_profile_custom_to_builtin() { + local dir=$1 + + setup $dir || return 1 + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + + # Verify the default mclock profile on the OSD + local def_mclock_profile + def_mclock_profile=$(ceph config get osd.0 osd_mclock_profile) + test "$def_mclock_profile" = "balanced" || return 1 + + # Verify the running mClock profile + local orig_mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile |\ + jq .osd_mclock_profile) + orig_mclock_profile=$(eval echo $orig_mclock_profile) + test $orig_mclock_profile = "high_recovery_ops" || return 1 + + # Change the mclock profile to 'custom' + ceph tell osd.0 config set osd_mclock_profile custom || return 1 + + # Verify that the mclock profile is set to 'custom' on the OSDs + local mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile | \ + jq .osd_mclock_profile) + mclock_profile=$(eval echo $mclock_profile) + test $mclock_profile = "custom" || return 1 + + # Save the original client reservations allocated to the OSDs + local client_res + client_res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + echo "Original client_res for osd.0 = $client_res" + + # Change a mclock config param and confirm the change + local client_res_new=$(echo "$client_res + 0.1" | bc -l) + echo "client_res_new = $client_res_new" + ceph config set osd osd_mclock_scheduler_client_res \ + $client_res_new || return 1 + # Check value in config monitor db + local res=$(ceph config get osd.0 \ + osd_mclock_scheduler_client_res) || return 1 + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + # Check value in the in-memory 'values' map + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + + # Switch the mclock profile back to the original built-in profile. + # The config subsystem prevents the overwrite of the changed QoS config + # option above i.e. osd_mclock_scheduler_client_res. This fact is verified + # before proceeding to remove the entry from the config monitor db. After + # the config entry is removed, the original value for the config option is + # restored and is verified. + ceph tell osd.0 config set osd_mclock_profile $orig_mclock_profile || return 1 + # Verify that the mclock profile is set back to the original on the OSD + eval mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile | \ + jq .osd_mclock_profile) + #mclock_profile=$(ceph config get osd.0 osd_mclock_profile) + test "$mclock_profile" = "$orig_mclock_profile" || return 1 + + # Verify that the new value is still in effect + # Check value in config monitor db + local res=$(ceph config get osd.0 \ + osd_mclock_scheduler_client_res) || return 1 + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + # Check value in the in-memory 'values' map + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + if (( $(echo "$res != $client_res_new" | bc -l) )); then + return 1 + fi + + # Remove the changed QoS config option from monitor db + ceph config rm osd osd_mclock_scheduler_client_res || return 1 + + sleep 5 # Allow time for change to take effect + + # Verify that the original values are now restored + # Check value in config monitor db + res=$(ceph config get osd.0 \ + osd_mclock_scheduler_client_res) || return 1 + if (( $(echo "$res != 0.0" | bc -l) )); then + return 1 + fi + + # Check value in the in-memory 'values' map + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get osd_mclock_scheduler_client_res | \ + jq .osd_mclock_scheduler_client_res | bc) + if (( $(echo "$res != $client_res" | bc -l) )); then + return 1 + fi + + teardown $dir || return 1 +} + +function TEST_recovery_limit_adjustment_mclock() { + local dir=$1 + + setup $dir || return 1 + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + local recoveries=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_recovery_max_active) + # Get default value + echo "$recoveries" | grep --quiet 'osd_recovery_max_active' || return 1 + + # Change the recovery limit without setting + # osd_mclock_override_recovery_settings option. Verify that the recovery + # limit is retained at its default value. + ceph config set osd.0 osd_recovery_max_active 10 || return 1 + sleep 2 # Allow time for change to take effect + local max_recoveries=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_recovery_max_active) + test "$max_recoveries" = "$recoveries" || return 1 + + # Change recovery limit after setting osd_mclock_override_recovery_settings. + # Verify that the recovery limit is modified. + ceph config set osd.0 osd_mclock_override_recovery_settings true || return 1 + ceph config set osd.0 osd_recovery_max_active 10 || return 1 + sleep 2 # Allow time for change to take effect + max_recoveries=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_recovery_max_active) + test "$max_recoveries" = '{"osd_recovery_max_active":"10"}' || return 1 + + teardown $dir || return 1 +} + +function TEST_backfill_limit_adjustment_mclock() { + local dir=$1 + + setup $dir || return 1 + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + local backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_max_backfills | jq .osd_max_backfills | bc) + # Get default value + echo "osd_max_backfills: $backfills" || return 1 + + # Change the backfill limit without setting + # osd_mclock_override_recovery_settings option. Verify that the backfill + # limit is retained at its default value. + ceph config set osd.0 osd_max_backfills 20 || return 1 + sleep 2 # Allow time for change to take effect + local max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_max_backfills | jq .osd_max_backfills | bc) + test $max_backfills = $backfills || return 1 + + # Verify local and async reserver settings are not changed + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .local_reservations.max_allowed | bc) + test $max_backfills = $backfills || return 1 + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .remote_reservations.max_allowed | bc) + test $max_backfills = $backfills || return 1 + + # Change backfills limit after setting osd_mclock_override_recovery_settings. + # Verify that the backfills limit is modified. + ceph config set osd.0 osd_mclock_override_recovery_settings true || return 1 + ceph config set osd.0 osd_max_backfills 20 || return 1 + sleep 2 # Allow time for change to take effect + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_max_backfills | jq .osd_max_backfills | bc) + test $max_backfills = 20 || return 1 + + # Verify local and async reserver settings are changed + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .local_reservations.max_allowed | bc) + test $max_backfills = 20 || return 1 + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .remote_reservations.max_allowed | bc) + test $max_backfills = 20 || return 1 + + # Kill osd and bring it back up. + # Confirm that the backfill settings are retained. + kill_daemons $dir TERM osd || return 1 + ceph osd down 0 || return 1 + wait_for_osd down 0 || return 1 + activate_osd $dir 0 --osd-op-queue=mclock_scheduler || return 1 + + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + config get osd_max_backfills | jq .osd_max_backfills | bc) + test $max_backfills = 20 || return 1 + + # Verify local and async reserver settings are changed + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .local_reservations.max_allowed | bc) + test $max_backfills = 20 || return 1 + max_backfills=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path osd.0) \ + dump_recovery_reservations | jq .remote_reservations.max_allowed | bc) + test $max_backfills = 20 || return 1 + + teardown $dir || return 1 +} + +function TEST_profile_disallow_builtin_params_modify() { + local dir=$1 + + setup $dir || return 1 + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + + # Verify that the default mclock profile is set on the OSD + local def_mclock_profile=$(ceph config get osd.0 osd_mclock_profile) + test "$def_mclock_profile" = "balanced" || return 1 + + # Verify the running mClock profile + local cur_mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile |\ + jq .osd_mclock_profile) + cur_mclock_profile=$(eval echo $cur_mclock_profile) + test $cur_mclock_profile = "high_recovery_ops" || return 1 + + declare -a options=("osd_mclock_scheduler_background_recovery_res" + "osd_mclock_scheduler_client_res") + + local retries=10 + local errors=0 + for opt in "${options[@]}" + do + # Try and change a mclock config param and confirm that no change occurred + local opt_val_orig=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get $opt | jq .$opt | bc) + local opt_val_new=$(echo "$opt_val_orig + 0.1" | bc -l) + ceph config set osd.0 $opt $opt_val_new || return 1 + + # Check configuration values + for count in $(seq 0 $(expr $retries - 1)) + do + errors=0 + sleep 2 # Allow time for changes to take effect + + echo "Check configuration values - Attempt#: $count" + # Check configuration value on Mon store (or the default) for the osd + local res=$(ceph config get osd.0 $opt) || return 1 + echo "Mon db (or default): osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check running configuration value using "config show" cmd + res=$(ceph config show osd.0 | grep $opt |\ + awk '{ print $2 }' | bc ) || return 1 + echo "Running config: osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) || \ + $(echo "$res != $opt_val_orig" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check value in the in-memory 'values' map is unmodified + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get $opt | jq .$opt | bc) + echo "Values map: osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) || \ + $(echo "$res != $opt_val_orig" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check if we succeeded or exhausted retry count + if [ $errors -eq 0 ] + then + break + elif [ $count -eq $(expr $retries - 1) ] + then + return 1 + fi + done + done + + teardown $dir || return 1 +} + +function TEST_profile_disallow_builtin_params_override() { + local dir=$1 + + setup $dir || return 1 + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + + run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1 + + # Verify that the default mclock profile is set on the OSD + local def_mclock_profile=$(ceph config get osd.0 osd_mclock_profile) + test "$def_mclock_profile" = "balanced" || return 1 + + # Verify the running mClock profile + local cur_mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get osd_mclock_profile |\ + jq .osd_mclock_profile) + cur_mclock_profile=$(eval echo $cur_mclock_profile) + test $cur_mclock_profile = "high_recovery_ops" || return 1 + + declare -a options=("osd_mclock_scheduler_background_recovery_res" + "osd_mclock_scheduler_client_res") + + local retries=10 + local errors=0 + for opt in "${options[@]}" + do + # Override a mclock config param and confirm that no change occurred + local opt_val_orig=$(CEPH_ARGS='' ceph --format=json daemon \ + $(get_asok_path osd.0) config get $opt | jq .$opt | bc) + local opt_val_new=$(echo "$opt_val_orig + 0.1" | bc -l) + ceph tell osd.0 config set $opt $opt_val_new || return 1 + + # Check configuration values + for count in $(seq 0 $(expr $retries - 1)) + do + errors=0 + sleep 2 # Allow time for changes to take effect + + echo "Check configuration values - Attempt#: $count" + # Check configuration value on Mon store (or the default) for the osd + local res=$(ceph config get osd.0 $opt) || return 1 + echo "Mon db (or default): osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check running configuration value using "config show" cmd + res=$(ceph config show osd.0 | grep $opt |\ + awk '{ print $2 }' | bc ) || return 1 + echo "Running config: osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) || \ + $(echo "$res != $opt_val_orig" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check value in the in-memory 'values' map is unmodified + res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \ + osd.0) config get $opt | jq .$opt | bc) + echo "Values map: osd.0 $opt = $res" + if (( $(echo "$res == $opt_val_new" | bc -l) || \ + $(echo "$res != $opt_val_orig" | bc -l) )); then + errors=$(expr $errors + 1) + fi + + # Check if we succeeded or exhausted retry count + if [ $errors -eq 0 ] + then + break + elif [ $count -eq $(expr $retries - 1) ] + then + return 1 + fi + done + done + + teardown $dir || return 1 +} + +main mclock-config "$@" + +# Local Variables: +# compile-command: "cd build ; make -j4 && \ +# ../qa/run-standalone.sh mclock-config.sh" +# End: diff --git a/qa/standalone/misc/network-ping.sh b/qa/standalone/misc/network-ping.sh new file mode 100755 index 000000000..4745108c5 --- /dev/null +++ b/qa/standalone/misc/network-ping.sh @@ -0,0 +1,169 @@ +#!/usr/bin/env bash + +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON="127.0.0.1:7146" # git grep '\<7146\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + CEPH_ARGS+="--mon-host=$CEPH_MON " + CEPH_ARGS+="--debug_disable_randomized_ping=true " + CEPH_ARGS+="--debug_heartbeat_testing_span=5 " + CEPH_ARGS+="--osd_heartbeat_interval=1 " + local funcs=${@:-$(set | sed -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} + for func in $funcs ; do + setup $dir || return 1 + $func $dir || return 1 + teardown $dir || return 1 + done +} + +function TEST_network_ping_test1() { + local dir=$1 + + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + + sleep 5 + + create_pool foo 16 + + # write some objects + timeout 20 rados bench -p foo 10 write -b 4096 --no-cleanup || return 1 + + # Get 1 cycle worth of ping data "1 minute" + sleep 10 + flush_pg_stats + + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "4" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "12" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + # Wait another 4 cycles to get "5 minute interval" + sleep 20 + flush_pg_stats + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "4" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "12" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + + # Wait another 10 cycles to get "15 minute interval" + sleep 50 + flush_pg_stats + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "0" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "1000" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "4" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network 0 | tee $dir/json + test "$(cat $dir/json | jq '.entries | length')" = "12" || return 1 + test "$(cat $dir/json | jq '.threshold')" = "0" || return 1 + + # Just check the threshold output matches the input + CEPH_ARGS='' ceph daemon $(get_asok_path mgr.x) dump_osd_network 99 | tee $dir/json + test "$(cat $dir/json | jq '.threshold')" = "99" || return 1 + CEPH_ARGS='' ceph daemon $(get_asok_path osd.0) dump_osd_network 98 | tee $dir/json + test "$(cat $dir/json | jq '.threshold')" = "98" || return 1 + + rm -f $dir/json +} + +# Test setting of mon_warn_on_slow_ping_time very low to +# get health warning +function TEST_network_ping_test2() { + local dir=$1 + + export CEPH_ARGS + export EXTRA_OPTS=" --mon_warn_on_slow_ping_time=0.001" + run_mon $dir a || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + + sleep 5 + ceph osd crush add-bucket dc1 datacenter + ceph osd crush add-bucket dc2 datacenter + ceph osd crush add-bucket dc3 datacenter + ceph osd crush add-bucket rack1 rack + ceph osd crush add-bucket rack2 rack + ceph osd crush add-bucket rack3 rack + ceph osd crush add-bucket host1 host + ceph osd crush add-bucket host2 host + ceph osd crush add-bucket host3 host + ceph osd crush move dc1 root=default + ceph osd crush move dc2 root=default + ceph osd crush move dc3 root=default + ceph osd crush move rack1 datacenter=dc1 + ceph osd crush move rack2 datacenter=dc2 + ceph osd crush move rack3 datacenter=dc3 + ceph osd crush move host1 rack=rack1 + ceph osd crush move host2 rack=rack2 + ceph osd crush move host3 rack=rack3 + ceph osd crush set osd.0 1.0 host=host1 + ceph osd crush set osd.1 1.0 host=host2 + ceph osd crush set osd.2 1.0 host=host3 + ceph osd crush rule create-simple myrule default host firstn + + create_pool foo 16 16 replicated myrule + + # write some objects + timeout 20 rados bench -p foo 10 write -b 4096 --no-cleanup || return 1 + + # Get at least 1 cycle of ping data (this test runs with 5 second cycles of 1 second pings) + sleep 10 + flush_pg_stats + + ceph health | tee $dir/health + grep -q "Slow OSD heartbeats" $dir/health || return 1 + + ceph health detail | tee $dir/health + grep -q "OSD_SLOW_PING_TIME_BACK" $dir/health || return 1 + grep -q "OSD_SLOW_PING_TIME_FRONT" $dir/health || return 1 + grep -q "Slow OSD heartbeats on front from osd[.][0-2] [[]dc[1-3],rack[1-3][]] \ +to osd[.][0-2] [[]dc[1-3],rack[1-3][]]" $dir/health || return 1 + rm -f $dir/health +} + +main network-ping "$@" + +# Local Variables: +# compile-command: "cd ../.. ; make -j4 && ../qa/run-standalone.sh network-ping.sh" +# End: diff --git a/qa/standalone/misc/ok-to-stop.sh b/qa/standalone/misc/ok-to-stop.sh new file mode 100755 index 000000000..dc9e7422f --- /dev/null +++ b/qa/standalone/misc/ok-to-stop.sh @@ -0,0 +1,296 @@ +#!/usr/bin/env bash + +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON_A="127.0.0.1:7150" # git grep '\<7150\>' : there must be only one + export CEPH_MON_B="127.0.0.1:7151" # git grep '\<7151\>' : there must be only one + export CEPH_MON_C="127.0.0.1:7152" # git grep '\<7152\>' : there must be only one + export CEPH_MON_D="127.0.0.1:7153" # git grep '\<7153\>' : there must be only one + export CEPH_MON_E="127.0.0.1:7154" # git grep '\<7154\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + export ORIG_CEPH_ARGS="$CEPH_ARGS" + + local funcs=${@:-$(set | ${SED} -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} + for func in $funcs ; do + setup $dir || return 1 + $func $dir || return 1 + kill_daemons $dir KILL || return 1 + teardown $dir || return 1 + done +} + +function TEST_1_mon_checks() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + + ceph mon ok-to-stop dne || return 1 + ! ceph mon ok-to-stop a || return 1 + + ! ceph mon ok-to-add-offline || return 1 + + ! ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm dne || return 1 +} + +function TEST_2_mons_checks() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B || return 1 + + ceph mon ok-to-stop dne || return 1 + ! ceph mon ok-to-stop a || return 1 + ! ceph mon ok-to-stop b || return 1 + ! ceph mon ok-to-stop a b || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm dne || return 1 +} + +function TEST_3_mons_checks() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B || return 1 + run_mon $dir c --public-addr=$CEPH_MON_C || return 1 + wait_for_quorum 60 3 + + ceph mon ok-to-stop dne || return 1 + ceph mon ok-to-stop a || return 1 + ceph mon ok-to-stop b || return 1 + ceph mon ok-to-stop c || return 1 + ! ceph mon ok-to-stop a b || return 1 + ! ceph mon ok-to-stop b c || return 1 + ! ceph mon ok-to-stop a b c || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm c || return 1 + + kill_daemons $dir KILL mon.b + wait_for_quorum 60 2 + + ! ceph mon ok-to-stop a || return 1 + ceph mon ok-to-stop b || return 1 + ! ceph mon ok-to-stop c || return 1 + + ! ceph mon ok-to-add-offline || return 1 + + ! ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ! ceph mon ok-to-rm c || return 1 +} + +function TEST_4_mons_checks() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C,$CEPH_MON_D " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B || return 1 + run_mon $dir c --public-addr=$CEPH_MON_C || return 1 + run_mon $dir d --public-addr=$CEPH_MON_D || return 1 + wait_for_quorum 60 4 + + ceph mon ok-to-stop dne || return 1 + ceph mon ok-to-stop a || return 1 + ceph mon ok-to-stop b || return 1 + ceph mon ok-to-stop c || return 1 + ceph mon ok-to-stop d || return 1 + ! ceph mon ok-to-stop a b || return 1 + ! ceph mon ok-to-stop c d || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm c || return 1 + + kill_daemons $dir KILL mon.a + wait_for_quorum 60 3 + + ceph mon ok-to-stop a || return 1 + ! ceph mon ok-to-stop b || return 1 + ! ceph mon ok-to-stop c || return 1 + ! ceph mon ok-to-stop d || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm c || return 1 + ceph mon ok-to-rm d || return 1 +} + +function TEST_5_mons_checks() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C,$CEPH_MON_D,$CEPH_MON_E " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B || return 1 + run_mon $dir c --public-addr=$CEPH_MON_C || return 1 + run_mon $dir d --public-addr=$CEPH_MON_D || return 1 + run_mon $dir e --public-addr=$CEPH_MON_E || return 1 + wait_for_quorum 60 5 + + ceph mon ok-to-stop dne || return 1 + ceph mon ok-to-stop a || return 1 + ceph mon ok-to-stop b || return 1 + ceph mon ok-to-stop c || return 1 + ceph mon ok-to-stop d || return 1 + ceph mon ok-to-stop e || return 1 + ceph mon ok-to-stop a b || return 1 + ceph mon ok-to-stop c d || return 1 + ! ceph mon ok-to-stop a b c || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm c || return 1 + ceph mon ok-to-rm d || return 1 + ceph mon ok-to-rm e || return 1 + + kill_daemons $dir KILL mon.a + wait_for_quorum 60 4 + + ceph mon ok-to-stop a || return 1 + ceph mon ok-to-stop b || return 1 + ceph mon ok-to-stop c || return 1 + ceph mon ok-to-stop d || return 1 + ceph mon ok-to-stop e || return 1 + + ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ceph mon ok-to-rm b || return 1 + ceph mon ok-to-rm c || return 1 + ceph mon ok-to-rm d || return 1 + ceph mon ok-to-rm e || return 1 + + kill_daemons $dir KILL mon.e + wait_for_quorum 60 3 + + ceph mon ok-to-stop a || return 1 + ! ceph mon ok-to-stop b || return 1 + ! ceph mon ok-to-stop c || return 1 + ! ceph mon ok-to-stop d || return 1 + ceph mon ok-to-stop e || return 1 + + ! ceph mon ok-to-add-offline || return 1 + + ceph mon ok-to-rm a || return 1 + ! ceph mon ok-to-rm b || return 1 + ! ceph mon ok-to-rm c || return 1 + ! ceph mon ok-to-rm d || return 1 + ceph mon ok-to-rm e || return 1 +} + +function TEST_0_mds() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 || return 1 + run_mds $dir a || return 1 + + ceph osd pool create meta 1 || return 1 + ceph osd pool create data 1 || return 1 + ceph fs new myfs meta data || return 1 + sleep 5 + + ! ceph mds ok-to-stop a || return 1 + ! ceph mds ok-to-stop a dne || return 1 + ceph mds ok-to-stop dne || return 1 + + run_mds $dir b || return 1 + sleep 5 + + ceph mds ok-to-stop a || return 1 + ceph mds ok-to-stop b || return 1 + ! ceph mds ok-to-stop a b || return 1 + ceph mds ok-to-stop a dne1 dne2 || return 1 + ceph mds ok-to-stop b dne || return 1 + ! ceph mds ok-to-stop a b dne || return 1 + ceph mds ok-to-stop dne1 dne2 || return 1 + + kill_daemons $dir KILL mds.a +} + +function TEST_0_osd() { + local dir=$1 + + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mgr $dir x || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + run_osd $dir 3 || return 1 + + ceph osd erasure-code-profile set ec-profile m=2 k=2 crush-failure-domain=osd || return 1 + ceph osd pool create ec erasure ec-profile || return 1 + + wait_for_clean || return 1 + + # with min_size 3, we can stop only 1 osd + ceph osd pool set ec min_size 3 || return 1 + wait_for_clean || return 1 + + ceph osd ok-to-stop 0 || return 1 + ceph osd ok-to-stop 1 || return 1 + ceph osd ok-to-stop 2 || return 1 + ceph osd ok-to-stop 3 || return 1 + ! ceph osd ok-to-stop 0 1 || return 1 + ! ceph osd ok-to-stop 2 3 || return 1 + ceph osd ok-to-stop 0 --max 2 | grep '[0]' || return 1 + ceph osd ok-to-stop 1 --max 2 | grep '[1]' || return 1 + + # with min_size 2 we can stop 1 osds + ceph osd pool set ec min_size 2 || return 1 + wait_for_clean || return 1 + + ceph osd ok-to-stop 0 1 || return 1 + ceph osd ok-to-stop 2 3 || return 1 + ! ceph osd ok-to-stop 0 1 2 || return 1 + ! ceph osd ok-to-stop 1 2 3 || return 1 + + ceph osd ok-to-stop 0 --max 2 | grep '[0,1]' || return 1 + ceph osd ok-to-stop 0 --max 20 | grep '[0,1]' || return 1 + ceph osd ok-to-stop 2 --max 2 | grep '[2,3]' || return 1 + ceph osd ok-to-stop 2 --max 20 | grep '[2,3]' || return 1 + + # we should get the same result with one of the osds already down + kill_daemons $dir TERM osd.0 || return 1 + ceph osd down 0 || return 1 + wait_for_peered || return 1 + + ceph osd ok-to-stop 0 || return 1 + ceph osd ok-to-stop 0 1 || return 1 + ! ceph osd ok-to-stop 0 1 2 || return 1 + ! ceph osd ok-to-stop 1 2 3 || return 1 +} + + +main ok-to-stop "$@" diff --git a/qa/standalone/misc/rados-striper.sh b/qa/standalone/misc/rados-striper.sh new file mode 100755 index 000000000..be6349b81 --- /dev/null +++ b/qa/standalone/misc/rados-striper.sh @@ -0,0 +1,101 @@ +#!/usr/bin/env bash +# +# Copyright (C) 2014 Red Hat <contact@redhat.com> +# +# Author: Sebastien Ponce <sebastien.ponce@cern.ch> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU Library Public License as published by +# the Free Software Foundation; either version 2, or (at your option) +# any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Library Public License for more details. +# +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON="127.0.0.1:7116" # git grep '\<7116\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + CEPH_ARGS+="--mon-host=$CEPH_MON " + + # setup + setup $dir || return 1 + + # create a cluster with one monitor and three osds + run_mon $dir a || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + create_rbd_pool || return 1 + + # create toyfile + dd if=/dev/urandom of=$dir/toyfile bs=1234 count=1 + + # put a striped object + rados --pool rbd --striper put toyfile $dir/toyfile || return 1 + + # stat it, with and without striping + rados --pool rbd --striper stat toyfile | cut -d ',' -f 2 > $dir/stripedStat || return 1 + rados --pool rbd stat toyfile.0000000000000000 | cut -d ',' -f 2 > $dir/stat || return 1 + echo ' size 1234' > $dir/refstat + diff -w $dir/stripedStat $dir/refstat || return 1 + diff -w $dir/stat $dir/refstat || return 1 + rados --pool rbd stat toyfile >& $dir/staterror + grep -q 'No such file or directory' $dir/staterror || return 1 + + # get the file back with and without striping + rados --pool rbd --striper get toyfile $dir/stripedGroup || return 1 + diff -w $dir/toyfile $dir/stripedGroup || return 1 + rados --pool rbd get toyfile.0000000000000000 $dir/nonSTripedGroup || return 1 + diff -w $dir/toyfile $dir/nonSTripedGroup || return 1 + + # test truncate + rados --pool rbd --striper truncate toyfile 12 + rados --pool rbd --striper stat toyfile | cut -d ',' -f 2 > $dir/stripedStat || return 1 + rados --pool rbd stat toyfile.0000000000000000 | cut -d ',' -f 2 > $dir/stat || return 1 + echo ' size 12' > $dir/reftrunc + diff -w $dir/stripedStat $dir/reftrunc || return 1 + diff -w $dir/stat $dir/reftrunc || return 1 + + # test xattrs + + rados --pool rbd --striper setxattr toyfile somexattr somevalue || return 1 + rados --pool rbd --striper getxattr toyfile somexattr > $dir/xattrvalue || return 1 + rados --pool rbd getxattr toyfile.0000000000000000 somexattr > $dir/xattrvalue2 || return 1 + echo 'somevalue' > $dir/refvalue + diff -w $dir/xattrvalue $dir/refvalue || return 1 + diff -w $dir/xattrvalue2 $dir/refvalue || return 1 + rados --pool rbd --striper listxattr toyfile > $dir/xattrlist || return 1 + echo 'somexattr' > $dir/reflist + diff -w $dir/xattrlist $dir/reflist || return 1 + rados --pool rbd listxattr toyfile.0000000000000000 | grep -v striper > $dir/xattrlist2 || return 1 + diff -w $dir/xattrlist2 $dir/reflist || return 1 + rados --pool rbd --striper rmxattr toyfile somexattr || return 1 + + local attr_not_found_str="No data available" + [ `uname` = FreeBSD ] && \ + attr_not_found_str="Attribute not found" + expect_failure $dir "$attr_not_found_str" \ + rados --pool rbd --striper getxattr toyfile somexattr || return 1 + expect_failure $dir "$attr_not_found_str" \ + rados --pool rbd getxattr toyfile.0000000000000000 somexattr || return 1 + + # test rm + rados --pool rbd --striper rm toyfile || return 1 + expect_failure $dir 'No such file or directory' \ + rados --pool rbd --striper stat toyfile || return 1 + expect_failure $dir 'No such file or directory' \ + rados --pool rbd stat toyfile.0000000000000000 || return 1 + + # cleanup + teardown $dir || return 1 +} + +main rados-striper "$@" diff --git a/qa/standalone/misc/test-ceph-helpers.sh b/qa/standalone/misc/test-ceph-helpers.sh new file mode 100755 index 000000000..e7805858a --- /dev/null +++ b/qa/standalone/misc/test-ceph-helpers.sh @@ -0,0 +1,21 @@ +#!/usr/bin/env bash +# +# Copyright (C) 2013,2014 Cloudwatt <libre.licensing@cloudwatt.com> +# Copyright (C) 2014 Red Hat <contact@redhat.com> +# Copyright (C) 2014 Federico Gimenez <fgimenez@coit.es> +# +# Author: Loic Dachary <loic@dachary.org> +# Author: Federico Gimenez <fgimenez@coit.es> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU Library Public License as published by +# the Free Software Foundation; either version 2, or (at your option) +# any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Library Public License for more details. +# + +$CEPH_ROOT/qa/standalone/ceph-helpers.sh TESTS "$@" diff --git a/qa/standalone/misc/test-snaptrim-stats.sh b/qa/standalone/misc/test-snaptrim-stats.sh new file mode 100755 index 000000000..98b3e4fdd --- /dev/null +++ b/qa/standalone/misc/test-snaptrim-stats.sh @@ -0,0 +1,188 @@ +#!/usr/bin/env bash +# +# Copyright (C) 2022 Red Hat <contact@redhat.com> +# +# Author: Sridhar Seshasayee <sseshasa@redhat.com> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU Library Public License as published by +# the Free Software Foundation; either version 2, or (at your option) +# any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Library Public License for more details. +# + +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON="127.0.0.1:7124" # git grep '\<7124\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + CEPH_ARGS+="--mon-host=$CEPH_MON " + CEPH_ARGS+="--debug-bluestore 20 " + + local funcs=${@:-$(set | sed -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} + for func in $funcs ; do + setup $dir || return 1 + $func $dir || return 1 + teardown $dir || return 1 + done +} + +function TEST_snaptrim_stats() { + local dir=$1 + local poolname=test + local OSDS=3 + local PGNUM=8 + local PGPNUM=8 + local objects=10 + local WAIT_FOR_UPDATE=10 + + setup $dir || return 1 + run_mon $dir a --osd_pool_default_size=$OSDS || return 1 + run_mgr $dir x || return 1 + for osd in $(seq 0 $(expr $OSDS - 1)) + do + run_osd $dir $osd --osd_pool_default_pg_autoscale_mode=off || return 1 + done + + # disable scrubs + ceph osd set noscrub || return 1 + ceph osd set nodeep-scrub || return 1 + + # Create a pool + create_pool $poolname $PGNUM $PGPNUM + wait_for_clean || return 1 + poolid=$(ceph osd dump | grep "^pool.*[']${poolname}[']" | awk '{ print $2 }') + + # write a few objects + TESTDATA="testdata.1" + dd if=/dev/urandom of=$TESTDATA bs=4096 count=1 + for i in `seq 1 $objects` + do + rados -p $poolname put obj${i} $TESTDATA + done + rm -f $TESTDATA + + # create a snapshot, clones + SNAP=1 + rados -p $poolname mksnap snap${SNAP} + TESTDATA="testdata.2" + dd if=/dev/urandom of=$TESTDATA bs=4096 count=1 + for i in `seq 1 $objects` + do + rados -p $poolname put obj${i} $TESTDATA + done + rm -f $TESTDATA + + # remove the snapshot, should trigger snaptrim + rados -p $poolname rmsnap snap${SNAP} + + # check for snaptrim stats + wait_for_clean || return 1 + sleep $WAIT_FOR_UPDATE + local objects_trimmed=0 + local snaptrim_duration_total=0.0 + for i in $(seq 0 $(expr $PGNUM - 1)) + do + local pgid="${poolid}.${i}" + objects_trimmed=$(expr $objects_trimmed + $(ceph pg $pgid query | \ + jq '.info.stats.objects_trimmed')) + snaptrim_duration_total=`echo $snaptrim_duration_total + $(ceph pg \ + $pgid query | jq '.info.stats.snaptrim_duration') | bc` + done + test $objects_trimmed -eq $objects || return 1 + echo "$snaptrim_duration_total > 0.0" | bc || return 1 + + teardown $dir || return 1 +} + +function TEST_snaptrim_stats_multiple_snaps() { + local dir=$1 + local poolname=test + local OSDS=3 + local PGNUM=8 + local PGPNUM=8 + local objects=10 + local WAIT_FOR_UPDATE=10 + + setup $dir || return 1 + run_mon $dir a --osd_pool_default_size=$OSDS || return 1 + run_mgr $dir x || return 1 + for osd in $(seq 0 $(expr $OSDS - 1)) + do + run_osd $dir $osd --osd_pool_default_pg_autoscale_mode=off || return 1 + done + + # disable scrubs + ceph osd set noscrub || return 1 + ceph osd set nodeep-scrub || return 1 + + # Create a pool + create_pool $poolname $PGNUM $PGPNUM + wait_for_clean || return 1 + poolid=$(ceph osd dump | grep "^pool.*[']${poolname}[']" | awk '{ print $2 }') + + # write a few objects + local TESTDATA="testdata.0" + dd if=/dev/urandom of=$TESTDATA bs=4096 count=1 + for i in `seq 1 $objects` + do + rados -p $poolname put obj${i} $TESTDATA + done + rm -f $TESTDATA + + # create snapshots, clones + NUMSNAPS=2 + for i in `seq 1 $NUMSNAPS` + do + rados -p $poolname mksnap snap${i} + TESTDATA="testdata".${i} + dd if=/dev/urandom of=$TESTDATA bs=4096 count=1 + for i in `seq 1 $objects` + do + rados -p $poolname put obj${i} $TESTDATA + done + rm -f $TESTDATA + done + + # remove the snapshots, should trigger snaptrim + local total_objects_trimmed=0 + for i in `seq 1 $NUMSNAPS` + do + rados -p $poolname rmsnap snap${i} + + # check for snaptrim stats + wait_for_clean || return 1 + sleep $WAIT_FOR_UPDATE + local objects_trimmed=0 + local snaptrim_duration_total=0.0 + for i in $(seq 0 $(expr $PGNUM - 1)) + do + local pgid="${poolid}.${i}" + objects_trimmed=$(expr $objects_trimmed + $(ceph pg $pgid query | \ + jq '.info.stats.objects_trimmed')) + snaptrim_duration_total=`echo $snaptrim_duration_total + $(ceph pg \ + $pgid query | jq '.info.stats.snaptrim_duration') | bc` + done + test $objects_trimmed -eq $objects || return 1 + echo "$snaptrim_duration_total > 0.0" | bc || return 1 + total_objects_trimmed=$(expr $total_objects_trimmed + $objects_trimmed) + done + + test $total_objects_trimmed -eq $((objects * NUMSNAPS)) || return 1 + + teardown $dir || return 1 +} +main test-snaptrim-stats "$@" + +# Local Variables: +# compile-command: "cd build ; make -j4 && \ +# ../qa/run-standalone.sh test-snaptrim-stats.sh" +# End: diff --git a/qa/standalone/misc/ver-health.sh b/qa/standalone/misc/ver-health.sh new file mode 100755 index 000000000..e03f8f4f5 --- /dev/null +++ b/qa/standalone/misc/ver-health.sh @@ -0,0 +1,231 @@ +#!/usr/bin/env bash +# +# Copyright (C) 2020 Red Hat <contact@redhat.com> +# +# Author: David Zafman <dzafman@redhat.com> +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU Library Public License as published by +# the Free Software Foundation; either version 2, or (at your option) +# any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU Library Public License for more details. +# +source $CEPH_ROOT/qa/standalone/ceph-helpers.sh + +function run() { + local dir=$1 + shift + + export CEPH_MON_A="127.0.0.1:7165" # git grep '\<7165\>' : there must be only one + export CEPH_MON_B="127.0.0.1:7166" # git grep '\<7166\>' : there must be only one + export CEPH_ARGS + CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " + CEPH_ARGS+="--mon-host=$CEPH_MON " + CEPH_ARGS+="--mon_health_to_clog_tick_interval=1.0 " + export ORIG_CEPH_ARGS="$CEPH_ARGS" + + local funcs=${@:-$(set | ${SED} -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} + for func in $funcs ; do + setup $dir || return 1 + $func $dir || return 1 + teardown $dir || return 1 + done +} + +function wait_for_health_string() { + local grep_string=$1 + local seconds=${2:-20} + + # Allow mon to notice version difference + set -o pipefail + PASSED="false" + for ((i=0; i < $seconds; i++)); do + if ceph health | grep -q "$grep_string" + then + PASSED="true" + break + fi + sleep 1 + done + set +o pipefail + + # Make sure health changed + if [ $PASSED = "false" ]; + then + return 1 + fi + return 0 +} + + + +# Test a single OSD with an old version and multiple OSDs with 2 different old versions +function TEST_check_version_health_1() { + local dir=$1 + + # Asssume MON_A is leader? + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + # setup + setup $dir || return 1 + + # create a cluster with two monitors and three osds + run_mon $dir a --public-addr=$CEPH_MON_A --mon_warn_older_version_delay=0 || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B --mon_warn_older_version_delay=0 || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + + sleep 5 + ceph health detail + # should not see this yet + ceph health detail | grep DAEMON_OLD_VERSION && return 1 + + kill_daemons $dir KILL osd.1 + ceph_debug_version_for_testing=01.00.00-gversion-test activate_osd $dir 1 + + wait_for_health_string "HEALTH_WARN .*There is a daemon running an older version of ceph" || return 1 + + ceph health detail + # Should notice that osd.1 is a different version + ceph health detail | grep -q "HEALTH_WARN .*There is a daemon running an older version of ceph" || return 1 + ceph health detail | grep -q "^[[]WRN[]] DAEMON_OLD_VERSION: There is a daemon running an older version of ceph" || return 1 + ceph health detail | grep -q "osd.1 is running an older version of ceph: 01.00.00-gversion-test" || return 1 + + kill_daemons $dir KILL osd.2 + ceph_debug_version_for_testing=01.00.00-gversion-test activate_osd $dir 2 + kill_daemons $dir KILL osd.0 + ceph_debug_version_for_testing=02.00.00-gversion-test activate_osd $dir 0 + + wait_for_health_string "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + + ceph health detail + ceph health detail | grep -q "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "^[[]ERR[]] DAEMON_OLD_VERSION: There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "osd.1 osd.2 are running an older version of ceph: 01.00.00-gversion-test" || return 1 + ceph health detail | grep -q "osd.0 is running an older version of ceph: 02.00.00-gversion-test" || return 1 +} + +# Test with 1 MON and 1 MDS with an older version, and add 2 OSDs with different versions +function TEST_check_version_health_2() { + local dir=$1 + + # Asssume MON_A is leader? + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + # setup + setup $dir || return 1 + + # create a cluster with all daemon types + run_mon $dir a --public-addr=$CEPH_MON_A --mon_warn_older_version_delay=0 || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B --mon_warn_older_version_delay=0 || return 1 + run_osd $dir 0 || return 1 + run_osd $dir 1 || return 1 + run_osd $dir 2 || return 1 + run_mgr $dir x || return 1 + run_mgr $dir y || return 1 + run_mds $dir m || return 1 + run_mds $dir n || return 1 + + sleep 5 + ceph health detail + # should not see this yet + ceph health detail | grep DAEMON_OLD_VERSION && return 1 + + kill_daemons $dir KILL mon.b + ceph_debug_version_for_testing=01.00.00-gversion-test run_mon $dir b --mon_warn_older_version_delay=0 + # XXX: Manager doesn't seem to use the test specific config for version + #kill_daemons $dir KILL mgr.x + #ceph_debug_version_for_testing=02.00.00-gversion-test run_mgr $dir x + kill_daemons $dir KILL mds.m + ceph_debug_version_for_testing=01.00.00-gversion-test run_mds $dir m + + wait_for_health_string "HEALTH_WARN .*There are daemons running an older version of ceph" || return 1 + + ceph health detail + # Should notice that mon.b and mds.m is a different version + ceph health detail | grep -q "HEALTH_WARN .*There are daemons running an older version of ceph" || return 1 + ceph health detail | grep -q "^[[]WRN[]] DAEMON_OLD_VERSION: There are daemons running an older version of ceph" || return 1 + ceph health detail | grep -q "mon.b mds.m are running an older version of ceph: 01.00.00-gversion-test" || return 1 + + kill_daemons $dir KILL osd.2 + ceph_debug_version_for_testing=01.00.00-gversion-test activate_osd $dir 2 + kill_daemons $dir KILL osd.0 + ceph_debug_version_for_testing=02.00.00-gversion-test activate_osd $dir 0 + + wait_for_health_string "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + + ceph health detail + ceph health | grep -q "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "^[[]ERR[]] DAEMON_OLD_VERSION: There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "mon.b osd.2 mds.m are running an older version of ceph: 01.00.00-gversion-test" || return 1 + ceph health detail | grep -q "osd.0 is running an older version of ceph: 02.00.00-gversion-test" || return 1 +} + +# Verify delay handling with same setup as test 1 +function TEST_check_version_health_3() { + local dir=$1 + + # Asssume MON_A is leader? + CEPH_ARGS="$ORIG_CEPH_ARGS --mon-host=$CEPH_MON_A " + # setup + setup $dir || return 1 + + # create a cluster with two monitors and three osds + run_mon $dir a --public-addr=$CEPH_MON_A || return 1 + run_mon $dir b --public-addr=$CEPH_MON_B || return 1 + + local start_osd_time=$SECONDS + # use memstore for faster bootup + EXTRA_OPTS=" --osd-objectstore=memstore" run_osd $dir 0 || return 1 + EXTRA_OPTS=" --osd-objectstore=memstore" run_osd $dir 1 || return 1 + EXTRA_OPTS=" --osd-objectstore=memstore" run_osd $dir 2 || return 1 + # take the time used for boot osds into consideration + local warn_older_version_delay=$(($SECONDS - $start_osd_time + 20)) + + sleep 5 + ceph health detail + # should not see this yet + ceph health detail | grep DAEMON_OLD_VERSION && return 1 + ceph tell 'mon.*' injectargs "--mon_warn_older_version_delay $warn_older_version_delay" + kill_daemons $dir KILL osd.1 + EXTRA_OPTS=" --osd-objectstore=memstore" \ + ceph_debug_version_for_testing=01.00.00-gversion-test \ + activate_osd $dir 1 + + # Wait 50% of 20 second delay config + sleep 10 + # should not see this yet + ceph health detail | grep DAEMON_OLD_VERSION && return 1 + + # Now make sure that at least 20 seconds have passed + wait_for_health_string "HEALTH_WARN .*There is a daemon running an older version of ceph" 20 || return 1 + + ceph health detail + # Should notice that osd.1 is a different version + ceph health detail | grep -q "HEALTH_WARN .*There is a daemon running an older version of ceph" || return 1 + ceph health detail | grep -q "^[[]WRN[]] DAEMON_OLD_VERSION: There is a daemon running an older version of ceph" || return 1 + ceph health detail | grep -q "osd.1 is running an older version of ceph: 01.00.00-gversion-test" || return 1 + + kill_daemons $dir KILL osd.2 + ceph_debug_version_for_testing=01.00.00-gversion-test activate_osd $dir 2 + kill_daemons $dir KILL osd.0 + ceph_debug_version_for_testing=02.00.00-gversion-test activate_osd $dir 0 + + wait_for_health_string "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + + ceph health detail + ceph health detail | grep -q "HEALTH_ERR .*There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "^[[]ERR[]] DAEMON_OLD_VERSION: There are daemons running multiple old versions of ceph" || return 1 + ceph health detail | grep -q "osd.1 osd.2 are running an older version of ceph: 01.00.00-gversion-test" || return 1 + ceph health detail | grep -q "osd.0 is running an older version of ceph: 02.00.00-gversion-test" || return 1 +} + +main ver-health "$@" + +# Local Variables: +# compile-command: "cd ../.. ; make -j4 && ../qa/run-standalone.sh ver-health.sh" +# End: |