mon/OSDMonitor: add 'osd tier add-cache <pool> <size>' command

This is a friendlier interface for setting up a cache tier with some
reasonable defaults (defined via config options).  This will simplify
the user experience and documentation.

Signed-off-by: Sage Weil <sage@inktank.com>

diff --git a/qa/workunits/cephtool/test.sh b/qa/workunits/cephtool/test.sh
index d4ccc9d178e4f538b1a181f7554ae0b8dc0c46c7..51e7ab4dbbe819c54d1e7bef5a4c45f87f5b6f51 100755 (executable)
--- a/qa/workunits/cephtool/test.sh
+++ b/qa/workunits/cephtool/test.sh
@@ -88,6 +88,13 @@ ceph osd tier remove data cache2
 ceph osd pool delete cache cache --yes-i-really-really-mean-it
 ceph osd pool delete cache2 cache2 --yes-i-really-really-mean-it
 
+# convenient add-cache command
+ceph osd pool create cache3 2
+ceph osd tier add-cache data cache3 1024000
+ceph osd dump | grep cache3 | grep bloom | grep 'false_positive_probability: 0.05' | grep 'target_bytes 1024000' | grep '1200s x4'
+ceph osd tier remove data cache3
+ceph osd pool delete cache3 cache3 --yes-i-really-really-mean-it
+
 # Assumes there are at least 3 MDSes and two OSDs
 #
 

diff --git a/src/common/config_opts.h b/src/common/config_opts.h
index 39a8341f6e199cd43d7bf5fb166795dfa8e30cf1..c5afcd4650f9da5f263a3351172f8850884137c4 100644 (file)
--- a/src/common/config_opts.h
+++ b/src/common/config_opts.h
@@ -431,6 +431,12 @@ OPTION(osd_pool_default_flag_hashpspool, OPT_BOOL, true)   // use new pg hashing
 OPTION(osd_pool_default_hit_set_bloom_fpp, OPT_FLOAT, .05)
 OPTION(osd_hit_set_min_size, OPT_INT, 1000)  // min target size for a HitSet
 OPTION(osd_hit_set_namespace, OPT_STR, ".ceph-internal") // rados namespace for hit_set tracking
+
+OPTION(osd_tier_default_cache_mode, OPT_STR, "writeback")
+OPTION(osd_tier_default_cache_hit_set_count, OPT_INT, 4)
+OPTION(osd_tier_default_cache_hit_set_period, OPT_INT, 1200)
+OPTION(osd_tier_default_cache_hit_set_type, OPT_STR, "bloom")
+
 OPTION(osd_map_dedup, OPT_BOOL, true)
 OPTION(osd_map_cache_size, OPT_INT, 500)
 OPTION(osd_map_message_max, OPT_INT, 100)  // max maps per MOSDMap message

diff --git a/src/mon/MonCommands.h b/src/mon/MonCommands.h
index 230bd1014b10bcf943a395c0494fb038d0762fa4..a6bf5e5791af6bd85e8e3f0dec226b65182d5544 100644 (file)
--- a/src/mon/MonCommands.h
+++ b/src/mon/MonCommands.h
@@ -571,6 +571,13 @@ COMMAND("osd tier remove-overlay " \
        "name=pool,type=CephPoolname ", \
        "remove the overlay pool for base pool <pool>", "osd", "rw", "cli,rest")
 
+COMMAND("osd tier add-cache " \
+       "name=pool,type=CephPoolname " \
+       "name=tierpool,type=CephPoolname " \
+       "name=size,type=CephInt,range=0", \
+       "add a cache <tierpool> of size <size> to existing pool <pool>", \
+       "osd", "rw", "cli,rest")
+
 /*
  * mon/ConfigKeyService.cc
  */

diff --git a/src/mon/OSDMonitor.cc b/src/mon/OSDMonitor.cc
index 5f61b4cf6e74c193591d0e1519f049f954d5c318..45c28f2378f53ca571ed5ef5d72c63adc659b0c6 100644 (file)
--- a/src/mon/OSDMonitor.cc
+++ b/src/mon/OSDMonitor.cc
@@ -4658,6 +4658,90 @@ done:
     wait_for_finished_proposal(new Monitor::C_Command(mon, m, 0, ss.str(),
                                              get_last_committed() + 1));
     return true;
+  } else if (prefix == "osd tier add-cache") {
+    string poolstr;
+    cmd_getval(g_ceph_context, cmdmap, "pool", poolstr);
+    int64_t pool_id = osdmap.lookup_pg_pool_name(poolstr);
+    if (pool_id < 0) {
+      ss << "unrecognized pool '" << poolstr << "'";
+      err = -ENOENT;
+      goto reply;
+    }
+    string tierpoolstr;
+    cmd_getval(g_ceph_context, cmdmap, "tierpool", tierpoolstr);
+    int64_t tierpool_id = osdmap.lookup_pg_pool_name(tierpoolstr);
+    if (tierpool_id < 0) {
+      ss << "unrecognized pool '" << tierpoolstr << "'";
+      err = -ENOENT;
+      goto reply;
+    }
+    const pg_pool_t *p = osdmap.get_pg_pool(pool_id);
+    assert(p);
+    const pg_pool_t *tp = osdmap.get_pg_pool(tierpool_id);
+    assert(tp);
+    if (p->tiers.count(tierpool_id)) {
+      assert(tp->tier_of == pool_id);
+      err = 0;
+      ss << "pool '" << tierpoolstr << "' is now (or already was) a tier of '" << poolstr << "'";
+      goto reply;
+    }
+    if (tp->is_tier()) {
+      ss << "tier pool '" << tierpoolstr << "' is already a tier of '"
+        << osdmap.get_pool_name(tp->tier_of) << "'";
+      err = -EINVAL;
+      goto reply;
+    }
+    int64_t size = 0;
+    cmd_getval(g_ceph_context, cmdmap, "size", size);
+    // make sure new tier is empty
+    const pool_stat_t& tier_stats =
+      mon->pgmon()->pg_map.get_pg_pool_sum_stat(tierpool_id);
+    if (tier_stats.stats.sum.num_objects != 0) {
+      ss << "tier pool '" << tierpoolstr << "' is not empty";
+      err = -ENOTEMPTY;
+      goto reply;
+    }
+    string modestr = g_conf->osd_tier_default_cache_mode;
+    pg_pool_t::cache_mode_t mode = pg_pool_t::get_cache_mode_from_str(modestr);
+    if (mode < 0) {
+      ss << "osd tier cache default mode '" << modestr << "' is not a valid cache mode";
+      err = -EINVAL;
+      goto reply;
+    }
+    HitSet::Params hsp;
+    if (g_conf->osd_tier_default_cache_hit_set_type == "bloom") {
+      BloomHitSet::Params *bsp = new BloomHitSet::Params;
+      bsp->set_fpp(g_conf->osd_pool_default_hit_set_bloom_fpp);
+      hsp = HitSet::Params(bsp);
+    } else if (g_conf->osd_tier_default_cache_hit_set_type == "explicit_hash") {
+      hsp = HitSet::Params(new ExplicitHashHitSet::Params);
+    }
+    else if (g_conf->osd_tier_default_cache_hit_set_type == "explicit_object") {
+      hsp = HitSet::Params(new ExplicitObjectHitSet::Params);
+    } else {
+      ss << "osd tier cache default hit set type '" <<
+       g_conf->osd_tier_default_cache_hit_set_type << "' is not a known type";
+      err = -EINVAL;
+      goto reply;
+    }
+    // go
+    pg_pool_t *np = pending_inc.get_new_pool(pool_id, p);
+    pg_pool_t *ntp = pending_inc.get_new_pool(tierpool_id, tp);
+    if (np->tiers.count(tierpool_id) || ntp->is_tier()) {
+      wait_for_finished_proposal(new C_RetryMessage(this, m));
+      return true;
+    }
+    np->tiers.insert(tierpool_id);
+    ntp->tier_of = pool_id;
+    ntp->cache_mode = mode;
+    ntp->hit_set_count = g_conf->osd_tier_default_cache_hit_set_count;
+    ntp->hit_set_period = g_conf->osd_tier_default_cache_hit_set_period;
+    ntp->hit_set_params = hsp;
+    ntp->target_max_bytes = size;
+    ss << "pool '" << tierpoolstr << "' is now (or already was) a cache tier of '" << poolstr << "'";
+    wait_for_finished_proposal(new Monitor::C_Command(mon, m, 0, ss.str(),
+                                             get_last_committed() + 1));
+    return true;
   } else if (prefix == "osd pool set-quota") {
     string poolstr;
     cmd_getval(g_ceph_context, cmdmap, "pool", poolstr);