Merge pull request #14490 from liewegas/wip-no-luminous

mon: add mon_debug_no_require_luminous

Reviewed-by: Kefu Chai <kchai@redhat.com>

qa/suites/rados/basic/z-require-luminous

@@ -0,0 +1 @@
+../thrash/z-require-luminous

qa/suites/rados/monthrash/z-require-luminous

@@ -0,0 +1 @@
+../thrash/z-require-luminous

qa/suites/rados/multimon/z-require-luminous

@@ -0,0 +1 @@
+../thrash/z-require-luminous

qa/suites/rados/thrash-erasure-code/z-require-luminous

@@ -0,0 +1 @@
+../thrash/z-require-luminous

qa/suites/rados/thrash/z-require-luminous/at-end.yaml

@@ -0,0 +1,14 @@
+# do not require luminous osds at mkfs time; only set flag at
+# the end of the test run, then do a final scrub (to convert any
+# legacy snapsets), and verify we are healthy.
+tasks:
+- exec_on_cleanup:
+    mon.a:
+      - ceph osd set require_luminous_osds
+overrides:
+  ceph:
+    conf:
+      global:
+        mon debug no require luminous: true
+  thrashosds:
+    chance_thrash_cluster_full: 0

qa/suites/rados/verify/z-require-luminous

@@ -0,0 +1 @@
+../thrash/z-require-luminous

qa/tasks/exec_on_cleanup.py

@@ -0,0 +1,62 @@
+"""
+Exececute custom commands during unwind/cleanup
+"""
+import logging
+import contextlib
+
+from teuthology import misc as teuthology
+from teuthology import contextutil
+
+log = logging.getLogger(__name__)
+
+@contextlib.contextmanager
+def task(ctx, config):
+    """
+    Execute commands on a given role
+
+        tasks:
+        - ceph:
+        - kclient: [client.a]
+        - exec:
+            client.a:
+              - "echo 'module libceph +p' > /sys/kernel/debug/dynamic_debug/control"
+              - "echo 'module ceph +p' > /sys/kernel/debug/dynamic_debug/control"
+        - interactive:
+
+    It stops and fails with the first command that does not return on success. It means
+    that if the first command fails, the second won't run at all.
+
+    To avoid confusion it is recommended to explicitly enclose the commands in 
+    double quotes. For instance if the command is false (without double quotes) it will
+    be interpreted as a boolean by the YAML parser.
+
+    :param ctx: Context
+    :param config: Configuration
+    """
+    try:
+        yield
+    finally:
+        log.info('Executing custom commands...')
+        assert isinstance(config, dict), "task exec got invalid config"
+
+        testdir = teuthology.get_testdir(ctx)
+
+        if 'all' in config and len(config) == 1:
+            a = config['all']
+            roles = teuthology.all_roles(ctx.cluster)
+            config = dict((id_, a) for id_ in roles)
+
+            for role, ls in config.iteritems():
+                (remote,) = ctx.cluster.only(role).remotes.iterkeys()
+                log.info('Running commands on role %s host %s', role, remote.name)
+                for c in ls:
+                    c.replace('$TESTDIR', testdir)
+                    remote.run(
+                        args=[
+                            'sudo',
+                            'TESTDIR={tdir}'.format(tdir=testdir),
+                            'bash',
+                            '-c',
+                            c],
+                    )
+

src/common/config_opts.h

@@ -370,6 +370,7 @@ OPTION(mon_debug_deprecated_as_obsolete, OPT_BOOL, false) // consider deprecated
 OPTION(mon_debug_dump_transactions, OPT_BOOL, false)
 OPTION(mon_debug_dump_json, OPT_BOOL, false)
 OPTION(mon_debug_dump_location, OPT_STR, "/var/log/ceph/$cluster-$name.tdump")
+OPTION(mon_debug_no_require_luminous, OPT_BOOL, false)
 OPTION(mon_inject_transaction_delay_max, OPT_DOUBLE, 10.0)      // seconds
 OPTION(mon_inject_transaction_delay_probability, OPT_DOUBLE, 0) // range [0, 1]
 

src/mon/OSDMonitor.cc

@@ -161,10 +161,11 @@ void OSDMonitor::create_initial()
   // new cluster should require latest by default
   newmap.set_flag(CEPH_OSDMAP_REQUIRE_JEWEL);
   newmap.set_flag(CEPH_OSDMAP_REQUIRE_KRAKEN);
-  newmap.set_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS);
-
-  newmap.full_ratio = g_conf->mon_osd_full_ratio;
-  newmap.nearfull_ratio = g_conf->mon_osd_nearfull_ratio;
+  if (!g_conf->mon_debug_no_require_luminous) {
+    newmap.set_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS);
+    newmap.full_ratio = g_conf->mon_osd_full_ratio;
+    newmap.nearfull_ratio = g_conf->mon_osd_nearfull_ratio;
+  }
 
   // encode into pending incremental
   newmap.encode(pending_inc.fullmap,
@@ -249,7 +250,9 @@ void OSDMonitor::update_from_paxos(bool *need_bootstrap)
     auto p = bl.begin();
     std::lock_guard<std::mutex> l(creating_pgs_lock);
     creating_pgs.decode(p);
-    dout(7) << __func__ << " loading creating_pgs e" << creating_pgs.last_scan_epoch << dendl;
+    dout(7) << __func__ << " loading creating_pgs last_scan_epoch "
+	    << creating_pgs.last_scan_epoch
+	    << " with " << creating_pgs.pgs.size() << " pgs" << dendl;
   }
 
   // walk through incrementals
@@ -1179,11 +1182,17 @@ void OSDMonitor::encode_pending(MonitorDBStore::TransactionRef t)
 void OSDMonitor::trim_creating_pgs(creating_pgs_t* creating_pgs,
 				   const PGMap& pgm)
 {
-  for (auto& pg : pgm.pg_stat) {
-    auto created = creating_pgs->pgs.find(pg.first);
-    if (created != creating_pgs->pgs.end()) {
-      creating_pgs->pgs.erase(created);
-      creating_pgs->created_pools.insert(pg.first.pool());
+  auto p = creating_pgs->pgs.begin();
+  while (p != creating_pgs->pgs.end()) {
+    auto q = pgm.pg_stat.find(p->first);
+    if (q != pgm.pg_stat.end() &&
+	!(q->second.state & PG_STATE_CREATING)) {
+      dout(20) << __func__ << " pgmap shows " << p->first << " is created"
+	       << dendl;
+      p = creating_pgs->pgs.erase(p);
+      creating_pgs->created_pools.insert(q->first.pool());
+    } else {
+      ++p;
     }
   }
 }
@@ -3459,7 +3468,9 @@ void OSDMonitor::get_health(list<pair<health_status_t,string> >& summary,
     }
 
     // warn about upgrade flags that can be set but are not.
-    if (HAVE_FEATURE(osdmap.get_up_osd_features(), SERVER_LUMINOUS) &&
+    if (g_conf->mon_debug_no_require_luminous) {
+      // ignore these checks
+    } else if (HAVE_FEATURE(osdmap.get_up_osd_features(), SERVER_LUMINOUS) &&
 	!osdmap.test_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS)) {
       string msg = "all OSDs are running luminous or later but the"
 	" 'require_luminous_osds' osdmap flag is not set";
@@ -7238,6 +7249,11 @@ bool OSDMonitor::prepare_command_impl(MonOpRequestRef op,
       err = -EPERM;
       goto reply;
     }
+    if (!osdmap.test_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS)) {
+      ss << "you must set the require_luminous_osds flag to use this feature";
+      err = -EPERM;
+      goto reply;
+    }
     err = check_cluster_features(CEPH_FEATUREMASK_OSDMAP_REMAP, ss);
     if (err == -EAGAIN)
       goto wait;
@@ -7293,6 +7309,11 @@ bool OSDMonitor::prepare_command_impl(MonOpRequestRef op,
       err = -EPERM;
       goto reply;
     }
+    if (!osdmap.test_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS)) {
+      ss << "you must set the require_luminous_osds flag to use this feature";
+      err = -EPERM;
+      goto reply;
+    }
     err = check_cluster_features(CEPH_FEATUREMASK_OSDMAP_REMAP, ss);
     if (err == -EAGAIN)
       goto wait;
@@ -7332,6 +7353,11 @@ bool OSDMonitor::prepare_command_impl(MonOpRequestRef op,
       err = -EPERM;
       goto reply;
     }
+    if (!osdmap.test_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS)) {
+      ss << "you must set the require_luminous_osds flag to use this feature";
+      err = -EPERM;
+      goto reply;
+    }
     err = check_cluster_features(CEPH_FEATUREMASK_OSDMAP_REMAP, ss);
     if (err == -EAGAIN)
       goto wait;
@@ -7399,6 +7425,11 @@ bool OSDMonitor::prepare_command_impl(MonOpRequestRef op,
       err = -EPERM;
       goto reply;
     }
+    if (!osdmap.test_flag(CEPH_OSDMAP_REQUIRE_LUMINOUS)) {
+      ss << "you must set the require_luminous_osds flag to use this feature";
+      err = -EPERM;
+      goto reply;
+    }
     err = check_cluster_features(CEPH_FEATUREMASK_OSDMAP_REMAP, ss);
     if (err == -EAGAIN)
       goto wait;