Merge pull request #5717: rgw: create a tool for orphaned objects cle…

…anup

Reviewed-by: Loic Dachary <ldachary@redhat.com>

src/CMakeLists.txt

@@ -762,7 +762,8 @@ if(${WITH_RADOSGW})
     rgw/rgw_main.cc)
 
   set(radosgw_admin_srcs
-    rgw/rgw_admin.cc)
+    rgw/rgw_admin.cc
+    rgw/rgw_orphan.cc)
 
   add_executable(radosgw ${radosgw_srcs} $<TARGET_OBJECTS:heap_profiler_objs>)
   target_link_libraries(radosgw rgw_a librados

src/common/config_opts.h

@@ -979,6 +979,7 @@ OPTION(rgw_op_thread_timeout, OPT_INT, 10*60)
 OPTION(rgw_op_thread_suicide_timeout, OPT_INT, 0)
 OPTION(rgw_thread_pool_size, OPT_INT, 100)
 OPTION(rgw_num_control_oids, OPT_INT, 8)
+OPTION(rgw_num_rados_handles, OPT_U32, 1)
 
 OPTION(rgw_zone, OPT_STR, "") // zone name
 OPTION(rgw_zone_root_pool, OPT_STR, ".rgw.root")    // pool where zone specific info is stored

src/rgw/Makefile.am

@@ -100,7 +100,7 @@ radosgw_CFLAGS = -I$(srcdir)/civetweb/include
 radosgw_LDADD = $(LIBRGW) $(LIBCIVETWEB) $(LIBRGW_DEPS) $(RESOLV_LIBS) $(CEPH_GLOBAL)
 bin_PROGRAMS += radosgw
 
-radosgw_admin_SOURCES = rgw/rgw_admin.cc
+radosgw_admin_SOURCES = rgw/rgw_admin.cc rgw/rgw_orphan.cc
 radosgw_admin_LDADD = $(LIBRGW) $(LIBRGW_DEPS) $(CEPH_GLOBAL)
 bin_PROGRAMS += radosgw-admin
 
@@ -141,6 +141,7 @@ noinst_HEADERS += \
 	rgw/rgw_metadata.h \
 	rgw/rgw_multi_del.h \
 	rgw/rgw_op.h \
+	rgw/rgw_orphan.h \
 	rgw/rgw_http_client.h \
 	rgw/rgw_swift.h \
 	rgw/rgw_swift_auth.h \

src/rgw/rgw_admin.cc

@@ -31,6 +31,7 @@ using namespace std;
 #include "rgw_formats.h"
 #include "rgw_usage.h"
 #include "rgw_replica_log.h"
+#include "rgw_orphan.h"
 
 #define dout_subsys ceph_subsys_rgw
 
@@ -232,6 +233,8 @@ enum {
   OPT_QUOTA_DISABLE,
   OPT_GC_LIST,
   OPT_GC_PROCESS,
+  OPT_ORPHANS_FIND,
+  OPT_ORPHANS_FINISH,
   OPT_REGION_GET,
   OPT_REGION_LIST,
   OPT_REGION_SET,
@@ -281,6 +284,7 @@ static int get_cmd(const char *cmd, const char *prev_cmd, bool *need_more)
       strcmp(cmd, "object") == 0 ||
       strcmp(cmd, "olh") == 0 ||
       strcmp(cmd, "opstate") == 0 ||
+      strcmp(cmd, "orphans") == 0 || 
       strcmp(cmd, "pool") == 0 ||
       strcmp(cmd, "pools") == 0 ||
       strcmp(cmd, "quota") == 0 ||
@@ -441,6 +445,11 @@ static int get_cmd(const char *cmd, const char *prev_cmd, bool *need_more)
       return OPT_GC_LIST;
     if (strcmp(cmd, "process") == 0)
       return OPT_GC_PROCESS;
+  } else if (strcmp(prev_cmd, "orphans") == 0) {
+    if (strcmp(cmd, "find") == 0)
+      return OPT_ORPHANS_FIND;
+    if (strcmp(cmd, "finish") == 0)
+      return OPT_ORPHANS_FINISH;
   } else if (strcmp(prev_cmd, "metadata") == 0) {
     if (strcmp(cmd, "get") == 0)
       return OPT_METADATA_GET;
@@ -1059,6 +1068,7 @@ int do_check_object_locator(const string& bucket_name, bool fix, bool remove_bad
   return 0;
 }
 
+
 int main(int argc, char **argv) 
 {
   vector<const char*> args;
@@ -1140,6 +1150,11 @@ int main(int argc, char **argv)
 
   BIIndexType bi_index_type = PlainIdx;
 
+  string job_id;
+  int num_shards = 0;
+  int max_concurrent_ios = 32;
+  uint64_t orphan_stale_secs = (24 * 3600);
+
   std::string val;
   std::ostringstream errs;
   string err;
@@ -1189,6 +1204,8 @@ int main(int argc, char **argv)
         cerr << "bad key type: " << key_type_str << std::endl;
         return usage();
       }
+    } else if (ceph_argparse_witharg(args, i, &val, "--job-id", (char*)NULL)) {
+      job_id = val;
     } else if (ceph_argparse_binary_flag(args, i, &gen_access_key, NULL, "--gen-access-key", (char*)NULL)) {
       // do nothing
     } else if (ceph_argparse_binary_flag(args, i, &gen_secret_key, NULL, "--gen-secret", (char*)NULL)) {
@@ -1238,6 +1255,12 @@ int main(int argc, char **argv)
       start_date = val;
     } else if (ceph_argparse_witharg(args, i, &val, "--end-date", "--end-time", (char*)NULL)) {
       end_date = val;
+    } else if (ceph_argparse_witharg(args, i, &val, "--num-shards", (char*)NULL)) {
+      num_shards = atoi(val.c_str());
+    } else if (ceph_argparse_witharg(args, i, &val, "--max-concurrent-ios", (char*)NULL)) {
+      max_concurrent_ios = atoi(val.c_str());
+    } else if (ceph_argparse_witharg(args, i, &val, "--orphan-stale-secs", (char*)NULL)) {
+      orphan_stale_secs = (uint64_t)atoi(val.c_str());
     } else if (ceph_argparse_witharg(args, i, &val, "--shard-id", (char*)NULL)) {
       shard_id = atoi(val.c_str());
       specified_shard_id = true;
@@ -2557,6 +2580,55 @@ int main(int argc, char **argv)
     }
   }
 
+  if (opt_cmd == OPT_ORPHANS_FIND) {
+    RGWOrphanSearch search(store, max_concurrent_ios, orphan_stale_secs);
+
+    if (job_id.empty()) {
+      cerr << "ERROR: --job-id not specified" << std::endl;
+      return EINVAL;
+    }
+    if (pool_name.empty()) {
+      cerr << "ERROR: --pool not specified" << std::endl;
+      return EINVAL;
+    }
+
+    RGWOrphanSearchInfo info;
+
+    info.pool = pool_name;
+    info.job_name = job_id;
+    info.num_shards = num_shards;
+
+    int ret = search.init(job_id, &info);
+    if (ret < 0) {
+      cerr << "could not init search, ret=" << ret << std::endl;
+      return -ret;
+    }
+    ret = search.run();
+    if (ret < 0) {
+      return -ret;
+    }
+  }
+
+  if (opt_cmd == OPT_ORPHANS_FINISH) {
+    RGWOrphanSearch search(store, max_concurrent_ios, orphan_stale_secs);
+
+    if (job_id.empty()) {
+      cerr << "ERROR: --job-id not specified" << std::endl;
+      return EINVAL;
+    }
+    int ret = search.init(job_id, NULL);
+    if (ret < 0) {
+      if (ret == -ENOENT) {
+        cerr << "job not found" << std::endl;
+      }
+      return -ret;
+    }
+    ret = search.finish();
+    if (ret < 0) {
+      return -ret;
+    }
+  }
+
   if (opt_cmd == OPT_USER_CHECK) {
     check_bad_user_bucket_mapping(store, user_id, fix);
   }

src/rgw/rgw_common.h

@@ -1316,31 +1316,46 @@ class rgw_obj {
    * part of the given namespace, it returns false.
    */
   static bool translate_raw_obj_to_obj_in_ns(string& obj, string& instance, string& ns) {
-    if (ns.empty()) {
-      if (obj[0] != '_')
-        return true;
-
-      if (obj.size() >= 2 && obj[1] == '_') {
-        obj = obj.substr(1);
+    if (obj[0] != '_') {
+      if (ns.empty()) {
         return true;
       }
-
       return false;
     }
 
-    if (obj[0] != '_' || obj.size() < 3) // for namespace, min size would be 3: _x_
+    string obj_ns;
+    bool ret = parse_raw_oid(obj, &obj, &instance, &obj_ns);
+    if (!ret) {
+      return ret;
+    }
+
+    return (ns == obj_ns);
+  }
+
+  static bool parse_raw_oid(const string& oid, string *obj_name, string *obj_instance, string *obj_ns) {
+    obj_instance->clear();
+    obj_ns->clear();
+    if (oid[0] != '_') {
+      *obj_name = oid;
+      return true;
+    }
+
+    if (oid.size() >= 2 && oid[1] == '_') {
+      *obj_name = oid.substr(1);
+      return true;
+    }
+
+    if (oid[0] != '_' || oid.size() < 3) // for namespace, min size would be 3: _x_
       return false;
 
-    int pos = obj.find('_', 1);
+    int pos = oid.find('_', 1);
     if (pos <= 1) // if it starts with __, it's not in our namespace
       return false;
 
-    string obj_ns = obj.substr(1, pos - 1);
-    parse_ns_field(obj_ns, instance);
-    if (obj_ns.compare(ns) != 0)
-        return false;
+    *obj_ns = oid.substr(1, pos - 1);
+    parse_ns_field(*obj_ns, *obj_instance);
 
-    obj = obj.substr(pos + 1);
+    *obj_name = oid.substr(pos + 1);
     return true;
   }
 

src/rgw/rgw_gc.cc

@@ -186,7 +186,7 @@ int RGWGC::process(int index, int max_secs)
         if (obj.pool != last_pool) {
           delete ctx;
           ctx = new IoCtx;
-	  ret = store->rados->ioctx_create(obj.pool.c_str(), *ctx);
+	  ret = store->get_rados_handle()->ioctx_create(obj.pool.c_str(), *ctx);
 	  if (ret < 0) {
 	    dout(0) << "ERROR: failed to create ioctx pool=" << obj.pool << dendl;
 	    continue;