[SERVER-78268] Remove shard slows down when there are multiple sharded collections and jumbo chunks Created: 21/Jun/23 Updated: 26/Oct/23 |
|
| Status: | Backlog |
| Project: | Core Server |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Major - P3 |
| Reporter: | Silvia Surroca | Assignee: | Backlog - Catalog and Routing |
| Resolution: | Unresolved | Votes: | 0 |
| Labels: | oldshardingemea, shardingemea-qw | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Catalog and Routing
|
||||||||
| Participants: | |||||||||
| Story Points: | 2 | ||||||||
| Description |
|
Context When the balancer is selecting the chunks to move, it goes through all the collections randomly. To prioritize the migrations from the draining shards, for every collection, we first check if the collection has chunks on a draining shard. What is wrong If a draining shard exists and that collection has chunks that can't be migrated from the draining shard (like jumbo chunks), that shard will be removed from the list of available shards to participate inĀ a migration during the current balancing round, which is not correct. That removal means that, during the current balancing round, the next collections won't have any chance to move their chunks from that shard. Therefore, the draining of a shard has been slowed down due to the addition of this line. |