From 0c7e59d2e09380bfa22e3695b74893bb93ce7882 Mon Sep 17 00:00:00 2001 From: Kamoltat Date: Tue, 13 Dec 2022 14:31:51 +0000 Subject: [PATCH] Revert "qa/standalone/mon: init mon-stretched-cluster.sh" This commit belongs to https://github.com/ceph/ceph/pull/48803 which introduced https://tracker.ceph.com/issues/58239. Therefore, we are reverting it. This reverts commit 025d3fa6b928d6fe2a794011e81db0c52d0c9cc0. Fixes: https://tracker.ceph.com/issues/58239 Signed-off-by: Kamoltat --- qa/standalone/mon/mon-stretched-cluster.sh | 149 --------------------- src/mon/Monitor.cc | 8 +- src/mon/OSDMonitor.cc | 4 - 3 files changed, 1 insertion(+), 160 deletions(-) delete mode 100755 qa/standalone/mon/mon-stretched-cluster.sh diff --git a/qa/standalone/mon/mon-stretched-cluster.sh b/qa/standalone/mon/mon-stretched-cluster.sh deleted file mode 100755 index b530f72942214..0000000000000 --- a/qa/standalone/mon/mon-stretched-cluster.sh +++ /dev/null @@ -1,149 +0,0 @@ -#!/usr/bin/env bash - -source $CEPH_ROOT/qa/standalone/ceph-helpers.sh -function run() { - local dir=$1 - shift - - export CEPH_MON_A="127.0.0.1:7139" # git grep '\<7139\>' : there must be only one - export CEPH_MON_B="127.0.0.1:7141" # git grep '\<7141\>' : there must be only one - export CEPH_MON_C="127.0.0.1:7142" # git grep '\<7142\>' : there must be only one - export CEPH_MON_D="127.0.0.1:7143" # git grep '\<7143\>' : there must be only one - export CEPH_MON_E="127.0.0.1:7144" # git grep '\<7144\>' : there must be only one - export CEPH_ARGS - CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " - - export BASE_CEPH_ARGS=$CEPH_ARGS - CEPH_ARGS+="--mon-host=$CEPH_MON_A" - - 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 -} -TEST_stretched_cluster_failover_add_three_osds(){ - local dir=$1 - local OSDS=8 - setup $dir || return 1 - - run_mon $dir a --public-addr $CEPH_MON_A || return 1 - wait_for_quorum 300 1 || return 1 - - run_mon $dir b --public-addr $CEPH_MON_B || return 1 - CEPH_ARGS="$BASE_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B" - wait_for_quorum 300 2 || return 1 - - run_mon $dir c --public-addr $CEPH_MON_C || return 1 - CEPH_ARGS="$BASE_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C" - wait_for_quorum 300 3 || return 1 - - run_mon $dir d --public-addr $CEPH_MON_D || return 1 - CEPH_ARGS="$BASE_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C,$CEPH_MON_D" - wait_for_quorum 300 4 || return 1 - - run_mon $dir e --public-addr $CEPH_MON_E || return 1 - CEPH_ARGS="$BASE_CEPH_ARGS --mon-host=$CEPH_MON_A,$CEPH_MON_B,$CEPH_MON_C,$CEPH_MON_D,$CEPH_MON_E" - wait_for_quorum 300 5 || return 1 - - ceph mon set election_strategy connectivity - ceph mon add disallowed_leader e - - run_mgr $dir x || return 1 - run_mgr $dir y || return 1 - run_mgr $dir z || return 1 - - for osd in $(seq 0 $(expr $OSDS - 1)) - do - run_osd $dir $osd || return 1 - done - - for zone in iris pze - do - ceph osd crush add-bucket $zone zone - ceph osd crush move $zone root=default - done - - - ceph osd crush add-bucket node-2 host - ceph osd crush add-bucket node-3 host - ceph osd crush add-bucket node-4 host - ceph osd crush add-bucket node-5 host - - ceph osd crush move node-2 zone=iris - ceph osd crush move node-3 zone=iris - ceph osd crush move node-4 zone=pze - ceph osd crush move node-5 zone=pze - - ceph osd crush move osd.0 host=node-2 - ceph osd crush move osd.1 host=node-2 - ceph osd crush move osd.2 host=node-3 - ceph osd crush move osd.3 host=node-3 - ceph osd crush move osd.4 host=node-4 - ceph osd crush move osd.5 host=node-4 - ceph osd crush move osd.6 host=node-5 - ceph osd crush move osd.7 host=node-5 - - ceph mon set_location a zone=iris host=node-2 - ceph mon set_location b zone=iris host=node-3 - ceph mon set_location c zone=pze host=node-4 - ceph mon set_location d zone=pze host=node-5 - - hostname=$(hostname -s) - ceph osd crush remove $hostname || return 1 - ceph osd getcrushmap > crushmap || return 1 - crushtool --decompile crushmap > crushmap.txt || return 1 - sed 's/^# end crush map$//' crushmap.txt > crushmap_modified.txt || return 1 - cat >> crushmap_modified.txt << EOF -rule stretch_rule { - id 1 - type replicated - min_size 1 - max_size 10 - step take iris - step chooseleaf firstn 2 type host - step emit - step take pze - step chooseleaf firstn 2 type host - step emit -} - -# end crush map -EOF - - crushtool --compile crushmap_modified.txt -o crushmap.bin || return 1 - ceph osd setcrushmap -i crushmap.bin || return 1 - local stretched_poolname=stretched_rbdpool - ceph osd pool create $stretched_poolname 32 32 stretch_rule || return 1 - ceph osd pool set $stretched_poolname size 4 || return 1 - - sleep 3 - - ceph mon set_location e zone=arbiter host=node-1 - ceph mon enable_stretch_mode e stretch_rule zone - - kill_daemons $dir KILL mon.c || return 1 - kill_daemons $dir KILL mon.d || return 1 - - kill_daemons $dir KILL osd.4 || return 1 - kill_daemons $dir KILL osd.5 || return 1 - kill_daemons $dir KILL osd.6 || return 1 - kill_daemons $dir KILL osd.7 || return 1 - - ceph -s - - sleep 3 - - run_osd $dir 8 || return 1 - run_osd $dir 9 || return 1 - run_osd $dir 10 || return 1 - - ceph -s - - sleep 3 - - teardown $dir || return 1 - -} -main mon-stretched-cluster "$@" \ No newline at end of file diff --git a/src/mon/Monitor.cc b/src/mon/Monitor.cc index 418436de2ccb6..e5f304d10b213 100644 --- a/src/mon/Monitor.cc +++ b/src/mon/Monitor.cc @@ -6582,7 +6582,6 @@ void Monitor::notify_new_monmap(bool can_change_external_state) if (is_stretch_mode()) { if (!monmap->stretch_marked_down_mons.empty()) { - dout(20) << __func__ << " stretch_marked_down_mons: " << monmap->stretch_marked_down_mons << dendl; set_degraded_stretch_mode(); } } @@ -6688,14 +6687,10 @@ struct CMonGoRecovery : public Context { void Monitor::go_recovery_stretch_mode() { dout(20) << __func__ << dendl; - dout(20) << "is_leader(): " << is_leader() << dendl; if (!is_leader()) return; - dout(20) << "is_degraded_stretch_mode(): " << is_degraded_stretch_mode() << dendl; if (!is_degraded_stretch_mode()) return; - dout(20) << "is_recovering_stretch_mode(): " << is_recovering_stretch_mode() << dendl; if (is_recovering_stretch_mode()) return; - dout(20) << "dead_mon_buckets.size(): " << dead_mon_buckets.size() << dendl; - dout(20) << "dead_mon_buckets: " << dead_mon_buckets << dendl; + if (dead_mon_buckets.size()) { ceph_assert( 0 == "how did we try and do stretch recovery while we have dead monitor buckets?"); // we can't recover if we are missing monitors in a zone! @@ -6776,7 +6771,6 @@ void Monitor::trigger_degraded_stretch_mode(const set& dead_mons, void Monitor::set_degraded_stretch_mode() { - dout(20) << __func__ << dendl; degraded_stretch_mode = true; recovering_stretch_mode = false; osdmon()->set_degraded_stretch_mode(); diff --git a/src/mon/OSDMonitor.cc b/src/mon/OSDMonitor.cc index 97034994546a6..3191ed5bf18bf 100644 --- a/src/mon/OSDMonitor.cc +++ b/src/mon/OSDMonitor.cc @@ -954,10 +954,6 @@ void OSDMonitor::update_from_paxos(bool *need_bootstrap) dout(20) << "Degraded stretch mode set in this map" << dendl; if (!osdmap.recovering_stretch_mode) { mon.set_degraded_stretch_mode(); - dout(20) << "prev_num_up_osd: " << prev_num_up_osd << dendl; - dout(20) << "osdmap.num_up_osd: " << osdmap.num_up_osd << dendl; - dout(20) << "osdmap.num_osd: " << osdmap.num_osd << dendl; - dout(20) << "mon_stretch_cluster_recovery_ratio: " << cct->_conf.get_val("mon_stretch_cluster_recovery_ratio") << dendl; if (prev_num_up_osd < osdmap.num_up_osd && (osdmap.num_up_osd / (double)osdmap.num_osd) > cct->_conf.get_val("mon_stretch_cluster_recovery_ratio")) { -- 2.39.5