ceph/doc/radosgw/dynamicresharding.rst

   1 .. _rgw_dynamic_bucket_index_resharding:
   2
   3 ===================================
   4 RGW Dynamic Bucket Index Resharding
   5 ===================================
   6
   7 .. versionadded:: Luminous
   8
   9 A large bucket index can lead to performance problems. In order
  10 to address this problem we introduced bucket index sharding.
  11 Until Luminous, changing the number of bucket shards (resharding)
  12 needed to be done offline. Starting with Luminous we support
  13 online bucket resharding.
  14
  15 Each bucket index shard can handle its entries efficiently up until
  16 reaching a certain threshold number of entries. If this threshold is
  17 exceeded the system can encounter performance issues. The dynamic
  18 resharding feature detects this situation and automatically increases
  19 the number of shards used by the bucket index, resulting in the
  20 reduction of the number of entries in each bucket index shard. This
  21 process is transparent to the user.
  22
  23 By default dynamic bucket index resharding can only increase the
  24 number of bucket index sharts to 1999, although the upper-bound is a
  25 configuration parameter (see Configuration below). Furthermore, when
  26 possible, the process chooses a prime number of bucket index shards to
  27 help spread the number of bucket index entries across the bucket index
  28 shards more evenly.
  29
  30 The detection process runs in a background process that periodically
  31 scans all the buckets. A bucket that requires resharding is added to
  32 the resharding queue and will be scheduled to be resharded later. The
  33 reshard thread runs in the background and execute the scheduled
  34 resharding tasks, one at a time.
  35
  36 Multisite
  37 =========
  38
  39 Dynamic resharding is not supported in a multisite environment.
  40
  41
  42 Configuration
  43 =============
  44
  45 Enable/Disable dynamic bucket index resharding:
  46
  47 - ``rgw_dynamic_resharding``:  true/false, default: true
  48
  49 Configuration options that control the resharding process:
  50
  51 - ``rgw_max_objs_per_shard``: maximum number of objects per bucket index shard before resharding is triggered, default: 100000 objects
  52
  53 - ``rgw_max_dynamic_shards``: maximum number of shards that dynamic bucket index resharding can increase to, default: 1999
  54
  55 - ``rgw_reshard_bucket_lock_duration``: duration, in seconds, of lock on bucket obj during resharding, default: 360 seconds (i.e., 6 minutes)
  56
  57 - ``rgw_reshard_thread_interval``: maximum time, in seconds, between rounds of resharding queue processing, default: 600 seconds (i.e., 10 minutes)
  58
  59 - ``rgw_reshard_num_logs``: number of shards for the resharding queue, default: 16
  60
  61 Admin commands
  62 ==============
  63
  64 Add a bucket to the resharding queue
  65 ------------------------------------
  66
  67 ::
  68
  69    # radosgw-admin reshard add --bucket <bucket_name> --num-shards <new number of shards>
  70
  71 List resharding queue
  72 ---------------------
  73
  74 ::
  75
  76    # radosgw-admin reshard list
  77
  78 Process tasks on the resharding queue
  79 -------------------------------------
  80
  81 ::
  82
  83    # radosgw-admin reshard process
  84
  85 Bucket resharding status
  86 ------------------------
  87
  88 ::
  89
  90    # radosgw-admin reshard status --bucket <bucket_name>
  91
  92 The output is a json array of 3 objects (reshard_status, new_bucket_instance_id, num_shards) per shard.
  93
  94 For example, the output at different Dynamic Resharding stages is shown below:
  95
  96 ``1. Before resharding occurred:``
  97 ::
  98
  99   [
 100     {
 101         "reshard_status": "not-resharding",
 102         "new_bucket_instance_id": "",
 103         "num_shards": -1
 104     }
 105   ]
 106
 107 ``2. During resharding:``
 108 ::
 109
 110   [
 111     {
 112         "reshard_status": "in-progress",
 113         "new_bucket_instance_id": "1179f470-2ebf-4630-8ec3-c9922da887fd.8652.1",
 114         "num_shards": 2
 115     },
 116     {
 117         "reshard_status": "in-progress",
 118         "new_bucket_instance_id": "1179f470-2ebf-4630-8ec3-c9922da887fd.8652.1",
 119         "num_shards": 2
 120     }
 121   ]
 122
 123 ``3, After resharding completed:``
 124 ::
 125
 126   [
 127     {
 128         "reshard_status": "not-resharding",
 129         "new_bucket_instance_id": "",
 130         "num_shards": -1
 131     },
 132     {
 133         "reshard_status": "not-resharding",
 134         "new_bucket_instance_id": "",
 135         "num_shards": -1
 136     }
 137   ]
 138
 139
 140 Cancel pending bucket resharding
 141 --------------------------------
 142
 143 Note: Ongoing bucket resharding operations cannot be cancelled. ::
 144
 145    # radosgw-admin reshard cancel --bucket <bucket_name>
 146
 147 Manual immediate bucket resharding
 148 ----------------------------------
 149
 150 ::
 151
 152    # radosgw-admin bucket reshard --bucket <bucket_name> --num-shards <new number of shards>
 153
 154 When choosing a number of shards, the administrator should keep a
 155 number of items in mind. Ideally the administrator is aiming for no
 156 more than 100000 entries per shard, now and through some future point
 157 in time.
 158
 159 Additionally, bucket index shards that are prime numbers tend to work
 160 better in evenly distributing bucket index entries across the
 161 shards. For example, 7001 bucket index shards is better than 7000
 162 since the former is prime. A variety of web sites have lists of prime
 163 numbers; search for "list of prime numbers" withy your favorite web
 164 search engine to locate some web sites.
 165
 166 Troubleshooting
 167 ===============
 168
 169 Clusters prior to Luminous 12.2.11 and Mimic 13.2.5 left behind stale bucket
 170 instance entries, which were not automatically cleaned up. The issue also affected
 171 LifeCycle policies, which were not applied to resharded buckets anymore. Both of
 172 these issues can be worked around using a couple of radosgw-admin commands.
 173
 174 Stale instance management
 175 -------------------------
 176
 177 List the stale instances in a cluster that are ready to be cleaned up.
 178
 179 ::
 180
 181    # radosgw-admin reshard stale-instances list
 182
 183 Clean up the stale instances in a cluster. Note: cleanup of these
 184 instances should only be done on a single site cluster.
 185
 186 ::
 187
 188    # radosgw-admin reshard stale-instances rm
 189
 190
 191 Lifecycle fixes
 192 ---------------
 193
 194 For clusters that had resharded instances, it is highly likely that the old
 195 lifecycle processes would have flagged and deleted lifecycle processing as the
 196 bucket instance changed during a reshard. While this is fixed for newer clusters
 197 (from Mimic 13.2.6 and Luminous 12.2.12), older buckets that had lifecycle policies and
 198 that have undergone resharding will have to be manually fixed.
 199
 200 The command to do so is:
 201
 202 ::
 203
 204    # radosgw-admin lc reshard fix --bucket {bucketname}
 205
 206
 207 As a convenience wrapper, if the ``--bucket`` argument is dropped then this
 208 command will try and fix lifecycle policies for all the buckets in the cluster.
 209
 210 Object Expirer fixes
 211 --------------------
 212
 213 Objects subject to Swift object expiration on older clusters may have
 214 been dropped from the log pool and never deleted after the bucket was
 215 resharded. This would happen if their expiration time was before the
 216 cluster was upgraded, but if their expiration was after the upgrade
 217 the objects would be correctly handled. To manage these expire-stale
 218 objects, radosgw-admin provides two subcommands.
 219
 220 Listing:
 221
 222 ::
 223
 224    # radosgw-admin objects expire-stale list --bucket {bucketname}
 225
 226 Displays a list of object names and expiration times in JSON format.
 227
 228 Deleting:
 229
 230 ::
 231
 232    # radosgw-admin objects expire-stale rm --bucket {bucketname}
 233
 234
 235 Initiates deletion of such objects, displaying a list of object names, expiration times, and deletion status in JSON format.