osd: ignore num_objects_dirty for old pools

Way back in a0ed9c20048750fd4b2c7ce0339fa8b20ef08ca3 we introduced the
dirty flag, but we did not track it in the stats until much later in
c561d5ea22c335e4c05fdc16ca8fb41f75f89d81.  Unfortunately this interval
spans the emperor release.  To avoid making scrub error out and require
repair on *any* of those old pools, flag stats that were encoded before
now such that the dirty stats are ignored.  Clear the flag if we *do*
do a repair so that it will be tracked properly thereafter.

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

diff --git a/src/osd/ReplicatedPG.cc b/src/osd/ReplicatedPG.cc
index 8ec8509217a943753607345d6cd471e1a5f27ca0..eb0a271006208b3138eaf6133a658a77910cf090 100644 (file)
--- a/src/osd/ReplicatedPG.cc
+++ b/src/osd/ReplicatedPG.cc
@@ -10024,12 +10024,14 @@ void ReplicatedPG::_scrub_finish()
   dout(10) << mode << " got "
           << scrub_cstat.sum.num_objects << "/" << info.stats.stats.sum.num_objects << " objects, "
           << scrub_cstat.sum.num_object_clones << "/" << info.stats.stats.sum.num_object_clones << " clones, "
+          << scrub_cstat.sum.num_objects_dirty << "/" << info.stats.stats.sum.num_objects_dirty << " dirty, "
           << scrub_cstat.sum.num_bytes << "/" << info.stats.stats.sum.num_bytes << " bytes."
           << dendl;
 
   if (scrub_cstat.sum.num_objects != info.stats.stats.sum.num_objects ||
       scrub_cstat.sum.num_object_clones != info.stats.stats.sum.num_object_clones ||
-      scrub_cstat.sum.num_objects_dirty != info.stats.stats.sum.num_objects_dirty ||
+      (scrub_cstat.sum.num_objects_dirty != info.stats.stats.sum.num_objects_dirty &&
+       !info.stats.dirty_stats_invalid) ||
       scrub_cstat.sum.num_whiteouts != info.stats.stats.sum.num_whiteouts ||
       scrub_cstat.sum.num_bytes != info.stats.stats.sum.num_bytes) {
     osd->clog.error() << info.pgid << " " << mode
@@ -10044,6 +10046,7 @@ void ReplicatedPG::_scrub_finish()
     if (repair) {
       ++scrubber.fixed;
       info.stats.stats = scrub_cstat;
+      info.stats.dirty_stats_invalid = false;
       publish_stats_to_osd();
       share_pg_info();
     }

diff --git a/src/osd/osd_types.cc b/src/osd/osd_types.cc
index 855b5d85c598f61b859e02b1a412ae2c1c74b807..d41958f4ce31833e2836c70d854df0a76a3300e5 100644 (file)
--- a/src/osd/osd_types.cc
+++ b/src/osd/osd_types.cc
@@ -1450,7 +1450,7 @@ void pg_stat_t::dump_brief(Formatter *f) const
 
 void pg_stat_t::encode(bufferlist &bl) const
 {
-  ENCODE_START(13, 8, bl);
+  ENCODE_START(14, 8, bl);
   ::encode(version, bl);
   ::encode(reported_seq, bl);
   ::encode(reported_epoch, bl);
@@ -1479,12 +1479,13 @@ void pg_stat_t::encode(bufferlist &bl) const
   ::encode(stats_invalid, bl);
   ::encode(last_clean_scrub_stamp, bl);
   ::encode(last_became_active, bl);
+  ::encode(dirty_stats_invalid, bl);
   ENCODE_FINISH(bl);
 }
 
 void pg_stat_t::decode(bufferlist::iterator &bl)
 {
-  DECODE_START_LEGACY_COMPAT_LEN(13, 8, 8, bl);
+  DECODE_START_LEGACY_COMPAT_LEN(14, 8, 8, bl);
   ::decode(version, bl);
   ::decode(reported_seq, bl);
   ::decode(reported_epoch, bl);
@@ -1564,6 +1565,13 @@ void pg_stat_t::decode(bufferlist::iterator &bl)
   } else {
     last_became_active = last_active;
   }
+  if (struct_v >= 14) {
+    ::decode(dirty_stats_invalid, bl);
+  } else {
+    // if we are decoding an old encoding of this object, then the
+    // encoder may not have supported num_objects_dirty accounting.
+    dirty_stats_invalid = true;
+  }
   DECODE_FINISH(bl);
 }
 

diff --git a/src/osd/osd_types.h b/src/osd/osd_types.h
index cf11aa1437e76d79b12abc7780dc0dbd7b17ae1c..71d5aad21307db8cc1f7626359eff354bbd62b9c 100644 (file)
--- a/src/osd/osd_types.h
+++ b/src/osd/osd_types.h
@@ -1151,6 +1151,10 @@ struct pg_stat_t {
 
   utime_t last_became_active;
 
+  /// true if num_objects_dirty is not accurate (because it was not
+  /// maintained starting from pool creation)
+  bool dirty_stats_invalid;
+
   pg_stat_t()
     : reported_seq(0),
       reported_epoch(0),
@@ -1159,7 +1163,8 @@ struct pg_stat_t {
       parent_split_bits(0),
       stats_invalid(false),
       log_size(0), ondisk_log_size(0),
-      mapping_epoch(0)
+      mapping_epoch(0),
+      dirty_stats_invalid(false)
   { }
 
   epoch_t get_effective_last_epoch_clean() const {