[SERVER-60894] Time Series collections bucket re-opening Created: 21/Oct/21  Updated: 06/Dec/22  Resolved: 15/Sep/22

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: New Feature Priority: Major - P3
Reporter: Michael Gargiulo Assignee: Backlog - Storage Execution Team
Resolution: Duplicate Votes: 0
Labels: time-series
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Duplicate
duplicates SERVER-66690 Insert into closed buckets identified... Closed
is duplicated by SERVER-61029 Inconsistent behavior for time series... Closed
Related
is related to SERVER-61045 Provide mechanism to compress existin... Backlog
Assigned Teams:
Storage Execution
Participants:

 Description   

When measurements are inserted significantly out-of-order, buckets tend to be closed often, which leads to very inefficient bucketing, and often buckets with only one measurement. Additionally, after a restart, we effectively close all buckets and have to start bucketing over again.

Currently, once a bucket is closed and removed from the BucketCatalog, it may not be re-opened. Instead, a bucket should only be permanently closed if it is full.

If there is no suitable open bucket for a measurement, we should attempt to find a closed bucket, load it into the BucketCatalog, and insert more measurements. 



 Comments   
Comment by Dan Larkin-York [ 09/Nov/21 ]

Changing this from 5.2 to 6.0 Required, as we decided to implement SERVER-61195 as a short-term workaround for 5.2.

Generated at Thu Feb 08 05:51:01 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.