This is a short python script who's purpose is to detect both long running and instantanious data loss on a high availibility system.
The idea is that you:
- run this script on an external client, writing data to the system
- do something terrible to the system that might cause DU (like reboot hardware)
- validate that the data written is still correct
This script takes into account that data that has not been acknowledged might be EITHER new or old data, therefore either is considered valid.
Use Cntrl+C (SIGINT) to kill this program gracefully.
returned error codes:
0: no errors1: write error, probably due to the device itself (DU)2: validation error, meaning corrupted data (DL)
Simple usage would just be:
curl https://git.io/dataloss -L > dataloss && python dataloss /path/to/folder -a
This will download the script and call it. The script will push data, waiting for a failure and constantly validating data
validate the data with:
python dataloss -v
usage: dataloss.py [-h] [-l LOG] [-v] [--bs BS] [--blocks BLOCKS] [-a]
[--period PERIOD] [--timeout TIMEOUT]
[--total-blocks TOTAL_BLOCKS]
path
Run some IO and be able to detect and validate failure
positional arguments:
path path to write-to or validate
optional arguments:
-h, --help show this help message and exit
-l LOG, --log LOG log to write or validate from.
default=/tmp/dataloss.log
-v, --validate validate using logfile provided
--bs BS block size default=4096
--blocks BLOCKS num of blocks before wrapping. default=1000
-a, --auto-validate if set, data will be validated while writting takes
place
--period PERIOD period between writes. default=0
--timeout TIMEOUT time in seconds to write. default=inf
--total-blocks TOTAL_BLOCKS
total number of blocks to write default=inf