Details
-
Bug
-
Resolution: Fixed
-
Critical
-
3.0
-
Security Level: Public
-
Build 3.0.0-385.
Platform = Physical
OS = CentOS 6.5
CPU = Intel Xeon E5-2680 v2
Memory = 256 GB
Disk = RAID 10 SSD
-
Untriaged
-
Centos 64-bit
-
2
-
Yes
-
June 30 - July 18
Description
I see similar problems in different test cases but in this ticket I want to focus on specific and quite extreme test case where I compare 2.5.0 and 3.0.0:
– 1 node, 1 bucket, 100M small items (~256 bytes), non-DGM
– 150K updates/sec front-end workload
– Minimal fragmentation threshold (2%).
Fragmentation and data usage is constantly growing. See attached graph and report for details.
Notice that disk persistence is way faster 3.0 so that compaction has more work to do.
However test environment is super fast, 40 cores, 256GB RAM, RAID 100 SSD. In fact, we utilize only 10% of disk resources and 25% of CPU resources.