-
Type:
Improvement
-
Resolution: Won't Do
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Catalog and Routing
Steps to repro:
- Start a data migration from shard-0 to an empty shard-1 using chunk migrations.
- run $shardedDataDistribution while the migration is ongoing
The $shardedDataDistribution output shows orphan count on shard-1. Orphans are documents not owned by a shard so it is probably showing the doc count that is not yet owned by shard-1. However, this is confusing because orphans are created on the donor shard so one would expect to only see orphans on shard-0.
It is worth investigating if we should show orphans only for donor shard.
db.aggregate([{$shardedDataDistribution: {}}, {$match: {ns: "testdb.biggie"}}] )
[
{
ns: 'testdb.biggie',
shards: [
{
shardName: 'atlas-2nqu8t-shard-0',
numOrphanedDocs: 128879333,
numOwnedDocuments: 832508835,
ownedSizeBytes: Long('963212722095'),
orphanedSizeBytes: Long('149113388281')
},
{
shardName: 'atlas-2nqu8t-shard-1',
numOrphanedDocs: 57690,
numOwnedDocuments: 145822363,
ownedSizeBytes: Long('168716473991'),
orphanedSizeBytes: 66747330
}
]
}
]
Atlas [mongos] admin> db.aggregate([{$shardedDataDistribution: {}}] )
[
{
ns: 'config.system.sessions',
shards: [
{
shardName: 'atlas-2nqu8t-shard-0',
numOrphanedDocs: 0,
numOwnedDocuments: 226,
ownedSizeBytes: 31188,
orphanedSizeBytes: 0
}
]
},
{
ns: 'testdb.biggie',
shards: [
{
shardName: 'atlas-2nqu8t-shard-0',
numOrphanedDocs: 128879333,
numOwnedDocuments: 832504886,
ownedSizeBytes: Long('963208153102'),
orphanedSizeBytes: Long('149113388281')
},
{
shardName: 'atlas-2nqu8t-shard-1',
numOrphanedDocs: 72113,
numOwnedDocuments: 145818651,
ownedSizeBytes: Long('168712179207'),
orphanedSizeBytes: 83434741
}
]
}
]
Orphan definition in our docs:
In a sharded cluster, orphaned documents are those documents on a shard that also exist in chunks on other shards. This is caused by a failed migration or an incomplete migration cleanup because of an atypical shutdown.
Orphaned documents are cleaned up automatically after a chunk migration completes.