distributed-system-analysis/multivm_db_bench

Name: multivm_db_bench

Owner: distributed-system-analysis

Description: Multi VM MariaDB benchmarking with sysbench for different AIO modes

Created: 2016-03-12 20:04:27.0

Updated: 2016-03-12 20:15:44.0

Pushed: 2016-05-09 11:21:19.0

Homepage:

Size: 79

Language: Shell

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

multivm_db_bench

Some automation scripts to use for running sysbench benchmarks for different thread counts on multiple Virtual Machines, for MariaDB.

NOTE: Switch to branch multi-client for a more generalistic approach of running sysbench with multiple clients, instead of VMs. Follow it's README for more help.

Usage

Refer to Prerequisites section first.

IMPORTANT

• Next time, you run automate_sysbench.sh after maybe changing AIO_MODE, sysbench/mariadb setup won't be installed again, but would only be cleaned up. If you wanna forcefully reinstall sysbench, be sure to update REINSTALL_OPTION=1 in multivm.config.

• ENABLE_PBENCH=1 by default in multivm.config. Change to 0 or empty string, if not needed.

• For entry point use (from within scripts/ folder):

age:
tomate_sysbench.sh <multivm.config path> <vm1> <vm2> <vm3>...

ample (use std output/error redirection to file since this runs for long)
tomate_sysbench.sh  multivm.config vm{1..8} >> sysbench.log 2>&1 &

• Display contents of results dir (check whether they start filling up..)

splay_results_dir_contents.sh <multivm.config path>

• Later, on completion (in an hour or so), use this to collect all results..

llect_sysbench_results.sh <multivm.config path>

NOTE:

• Currently some pssh methods might be commented out in multivm_setup_initiate.py. Use as per requirement..
• Also, if you're using parallel-ssh==0.80.7 python2 package less than 0.90, you might face ascii decoding related error.

  • Just edit the file: /usr/lib/python2.7/site-packages/pssh/ssh_client.py (as shown in traceback), and change the 'ascii' part to 'utf-8' as shown below:

    ange this: output = line.strip().decode('ascii')
     this: output = line.strip().decode('utf-8')
    

    ..and run again. This is incase, you're uanble to upgrade to 0.90 (maybe it's not available on pip yet) or any other reason.

PREREQUISITES

1. Before starting, if needed, change the mysql password in config file my.cnf.example as under the following section/parameter:

   ent]
   =root
   word=90feet-
   

2. Ensure that following major params are present corretly assigned under multivm.config:

   MODE='native'
   _TABLE_SIZE=1000000
   

   Check other params as per need.

3. Before running, ensure all files in the same directory as automate_sysbench.sh, i.e., you're supposed to run this from under scripts/ folder in this repo. This is until there's a packaged release.

4. The VM(s) should be up and running, and have the folders already mounted, as per the aio modes.

   • /home/native with aio=native
   • /home/threads with aio=threads

5. Install the python2 module parallel-ssh via pip, on your host. (you'd have to temporarily enable epel repo for installing pip on rhel)

6. Ensure you have passwordless ssh access to all VMs from host machine.