os/bluestore: update freelist on initial commit

It does not matter if we update the freelist in the initial commit or when
cleaning up the deferred transaction; both will eventually update the
persistent kv freelist.  We maintain one case to ensure that legacy
deferred events (from a kraken upgrade) release when they are replayed.

What matters while online is the Allocator, which has an independent
in-memory copy of the freelist to make decisions.  And we can delay that
as long as we want.  To avoid any concerns about deferred writes racing
against released blocks, just defer any release until the txc is fully
completed (including any deferred writes).  This ensures that even if we
have a pattern like

 txc 1: schedule deferred write on block A
 txc 2: release block A
 txc 1+2: commit
 txc 2: done!
 txc 1: do deferred write
 txc 1: done!

then txc 2 won't do its release because it is stuck behind txc 1 in the
OpSequencer queue:

 ...
 txc 1: reaped
 txc 2: reaped (and extents released to alloc)

This builds in some delay in just-released space being usable again, but
it should be a very small amount of space relative to the size of the
store!

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

diff --git a/src/os/bluestore/BlueStore.cc b/src/os/bluestore/BlueStore.cc
index 7dc49daf9571ce4a1137973abb5fa07450240426..03bb73528aa5ce193967775196ff0084d5a5040d 100644 (file)
--- a/src/os/bluestore/BlueStore.cc
+++ b/src/os/bluestore/BlueStore.cc
@@ -7430,6 +7430,11 @@ void BlueStore::_osr_reap_done(OpSequencer *osr)
         break;
       }
 
+      // release to allocator only after all preceding txc's have also
+      // finished any deferred writes that potentially land in these
+      // blocks
+      _txc_release_alloc(txc);
+
       if (!c && txc->first_collection) {
         c = txc->first_collection;
       }
@@ -7601,12 +7606,15 @@ void BlueStore::_kv_sync_thread()
       }
 
       // cleanup sync deferred keys
-      for (std::deque<TransContext *>::iterator it = deferred_cleaning.begin();
-           it != deferred_cleaning.end();
-           ++it) {
-       bluestore_deferred_transaction_t& wt =*(*it)->deferred_txn;
-       // kv metadata updates
-       _txc_finalize_kv(*it, synct);
+      for (auto txc : deferred_cleaning) {
+       bluestore_deferred_transaction_t& wt = *txc->deferred_txn;
+       if (!wt.released.empty()) {
+         // kraken replay compat only
+         txc->released = wt.released;
+         dout(10) << __func__ << " deferred txn has released " << txc->released
+                  << " (we just upgraded from kraken) on " << txc << dendl;
+         _txc_finalize_kv(txc, synct);
+       }
        // cleanup the deferred
        string key;
        get_deferred_key(wt.seq, &key);
@@ -7634,13 +7642,11 @@ void BlueStore::_kv_sync_thread()
       while (!kv_committing.empty()) {
        TransContext *txc = kv_committing.front();
        assert(txc->state == TransContext::STATE_KV_SUBMITTED);
-       _txc_release_alloc(txc);
        _txc_state_proc(txc);
        kv_committing.pop_front();
       }
       while (!deferred_cleaning.empty()) {
        TransContext *txc = deferred_cleaning.front();
-       _txc_release_alloc(txc);
        _txc_state_proc(txc);
        deferred_cleaning.pop_front();
       }
@@ -7711,11 +7717,6 @@ int BlueStore::_deferred_finish(TransContext *txc)
 {
   bluestore_deferred_transaction_t& wt = *txc->deferred_txn;
   dout(20) << __func__ << " txc " << " seq " << wt.seq << txc << dendl;
-
-  // move released back to txc
-  txc->deferred_txn->released.swap(txc->released);
-  assert(txc->deferred_txn->released.empty());
-
   std::lock_guard<std::mutex> l2(txc->osr->qlock);
   std::lock_guard<std::mutex> l(kv_lock);
   txc->state = TransContext::STATE_DEFERRED_CLEANUP;
@@ -7838,10 +7839,6 @@ int BlueStore::queue_transactions(
 
   // journal deferred items
   if (txc->deferred_txn) {
-    // move releases to after deferred
-    txc->deferred_txn->released.swap(txc->released);
-    assert(txc->released.empty());
-
     txc->deferred_txn->seq = ++deferred_seq;
     bufferlist bl;
     ::encode(*txc->deferred_txn, bl);