YAML parser will receive a test yaml file as an input.
A test yaml file should have the following parameters based on workloads to run:
- TEST_ID
- object_size
- part_size
- range_read
- part_range
- write_percentage
- read_percentage
- delete_percentage
- cleanup_percentage
- total_storage_size
- background_delete
- min_runtime
- sessions_per_node
- sessions
- tool
- operation
TEST_ID can be test ticket which is used for JIRA update or any unique string which is used to create log file for searching through run logs.
object_size can be given as fixed values, list of values and range of values for test workload, Range for object size can be mentioned using start and end.
- start
- end
part_size can be given as fixed values and range of values for multipart workloads, so those can have the following keys in it.
- start
- end
Start and End range parameters are object_size/part_size for which test will run. Sizes can be given from bytes, KB (1000 Bytes), MB (1000 KB) up to TB (1000 GB). We can also use KiB (1024 Bytes) format as well.
sessions_per_node multiplied with number_of_nodes (command line argument '-nn') will be treated as total number of sessions for a test workload. Default value of number_of_nodes is 1.
sessions specified in YAML file will be directly used to create total sessions.
range_read key is used to perform range read operations in a workload.
part_range is range of numbers from which random number of parts for multipart workloads can be calculated.
write_percentage is percentage of data to fill the storage size.
read_percentage is percentage of data to read from given storage size.
delete_percentage is percentage of data to be deleted from storage.
cleanup_percentage is Percentage(less or equal to write) of data once reached then do cleanup.
total_storage_size Total available storage from cluster or user input size will be used (MB, Mib to PB, Pib and if None will be fetched from cluster).
background_delete Enable/Disable background delete(Cortx specific)
min_runtime can be specified from seconds up to days. For example: 1d1h, 1h or 2d1h2s.
tool can be specified from one of these s3api, s3bench or warp. note:- s3bench and warp support to be added.
operation should be specified according to type of workload and need to map it in corio.py with appropriate test script.
total_samples Total number of samples given by user to define object size distribution.
Sample YAML file can be found at sample file
YAML files for various s3 workloads can be found at workload/s3/s3api/