[1.3][benchmark] Checking the impact on performance when a node is killed

[yosukehara]

Environment and Configuration

• OS: Ubuntu Server 14.04.3
• Erlang/OTP: 17.5
• LeoFS: 1.3.0
• CPU: Intel Xeon E5-2630 v3 @ 2.40GHz
• LeoFS' system
  • number of leo_gateway: 1
  • number of leo_storage: 5
  • number of leo_manager: 2

Report

• https://github.com/leo-project/notes/tree/master/leofs/benchmark/leofs/1.3/1.3.0/front/20161007_image_f4m_rw_nodekill_1

[yosukehara]

Everyone cannot understand the detail of the benchmark report. We need to add some items in the benchmark report as below:

### Summary
- Description

### Finding issues
- Subject
   - What happened and How to reproduce if possible
   - Information
       - `leofs-doctor` result
       - Summary of errors
           - Description
           - Error log files
       - [Status:[open|close]] [A link of LeoFS' issue]

### Next Action
- Subject
    - Description

[windkit]

[PUT] Node Overload, [GET] Stable

https://github.com/leo-project/notes/tree/master/leofs/benchmark/leofs/1.3/1.3.0/front/20161007_image_f4m_rw_nodekill_1

LeoFS Settings

• 1 gateway, 5 storage, 10Gbps Link
• No GW Cache
• SSD-backed Storage

Methodology

• Basho Bench leofs driver
• 64 processes in erlang performing HTTP PUT/GET
• Data Set: 4,000,000 x (4KB~128KB, avg. 32KB)
• Node Kill during Load/Read
• [PUT] Partitioned Sequential
• [GET] Random

Summary

• Loading Results with and without node being kiiled are similar
• During read, the received throughput from client is stable
  • Before the first node is killed, the effective data set is small enough to sit in memory (page cache)
  • After nodes are killed, nodes are receiving more requests

Next Action

• Similar to base image case [1.3][benchmark] Small size object benchmark #18, need a closer look on loading phase