a list of config files that would be used
default: N/A
backend that will be used, aws/gce/docker
default: N/A
class.method used to run the test. Filled automatically with run-test sct command.
default: N/A
Test duration (min). Parameter used to keep instances produced by tests
and for jenkins pipeline timeout and TimoutThread.
default: 60
Time in minutes, which is required to run prepare stress commands
defined in prepare_*_cmd for dataset generation, and is used in
test duration calculation
default: 300
Time in minutes, Time of execution for stress commands from stress_cmd parameters
and is used in test duration calculation
default: N/A
Number list of database data nodes in multiple data centers. To use with
multi data centers and zero nodes, dc with zero-nodes only should be set as 0,
ex. "3 3 0".
default: N/A
Number list of oracle test nodes in multiple data centers.
default: 1
Number list of loader nodes in multiple data centers
default: N/A
Number list of monitor nodes in multiple data centers
default: N/A
If True, all communication between nodes are via public addresses
default: N/A
The snitch class scylla would use
'GossipingPropertyFileSnitch' - default
'Ec2MultiRegionSnitch' - default on aws backend
'GoogleCloudSnitch'
default: N/A
Path to your user credentials. qa key are downloaded automatically from S3 bucket
default: N/A
Path to your user credentials. qa key are downloaded automatically from S3 bucket
default: N/A
scylla cloud cluster id
default: N/A
scylla cloud promproxy bearer_token to federate monitoring data into our monitoring instance
default: N/A
scylla cloud promproxy path to federate monitoring data into our monitoring instance
default: N/A
scylla cloud promproxy hostname to federate monitoring data into our monitoring instance
default: N/A
Type of IP used to connect to machine instances.
This depends on whether you are running your tests from a machine inside
your cloud provider, where it makes sense to use 'private', or outside (use 'public')
Default: Use public IPs to connect to instances (public)
Use private IPs to connect to instances (private)
Use IPv6 IPs to connect to instances (ipv6)
default: private
Url to the repo of scylla version to install scylla. Can provide specific version after a colon e.g: https://s3.amazonaws.com/downloads.scylladb.com/deb/ubuntu/scylla-2021.1.list:2021.1.18
default: N/A
APT keys for ScyllaDB repos
default: ['17723034C56D4B19', '5E08FBD8B5D6EC9C', 'D0A112E067426AB2', '491C93B9DE7496A7', 'A43E06657BAC99E3']
Url to the unified package of scylla version to install scylla
default: N/A
Install Scylla without required root priviledge
default: N/A
Scylla install mode, repo/offline/web
default: repo
Version of scylla to install, ex. '2.3.1'
Automatically lookup AMIs and repo links for formal versions.
WARNING: can't be used together with 'scylla_repo' or 'ami_id_db_scylla'
default: N/A
Format version of the user-data to use for scylla images,
default to what tagged on the image used
default: N/A
Format version of the user-data to use for scylla images,
default to what tagged on the image used
default: N/A
Version of scylla to use as oracle cluster with gemini tests, ex. '3.0.11'
Automatically lookup AMIs for formal versions.
WARNING: can't be used together with 'ami_id_db_oracle'
default: 2022.1.14
The distro name and family name to use. Example: 'ubuntu-jammy' or 'debian-bookworm'.
default: ubuntu-focal
The distro name and family name to use. Example: 'ubuntu-jammy' or 'debian-bookworm'.
default: ubuntu-jammy
List of distro features relevant to SCT test. Example: 'fips'.
default: N/A
Url to the repo of scylla version to install scylla from for managment tests
default: N/A
Url to the repo of scylla version to install c-s for loader
default: https://s3.amazonaws.com/downloads.scylladb.com/deb/ubuntu/scylla-5.2.list
Url to the repo of scylla manager version to install for management tests
default: N/A
Url to the repo of scylla manager agent version to install for management tests
default: N/A
Branch of scylla manager server and agent to install. Options in defaults/manager_versions.yaml
default: 3.4
Branch of scylla manager server and agent to upgrade to. Options in defaults/manager_versions.yaml
default: N/A
Branch of scylla db enterprise to install. Options in defaults/manager_versions.yaml
default: 2024
default: 3.4.0
Url to the scylla manager packages to install for management tests
default: N/A
Stress command for LWT performance test for INSERT baseline
default: N/A
Stress command for LWT performance test for DELETE baseline
default: N/A
Stress command for LWT performance test for UPDATE baseline
default: N/A
Stress command for LWT performance test for INSERT with IF NOT EXISTS
default: N/A
Stress command for LWT performance test for UPDATE with IF
default: N/A
Stress command for LWT performance test for UPDATE with IF EXISTS
default: N/A
Stress command for LWT performance test for DELETE with IF EXISTS
default: N/A
Stress command for LWT performance test for DELETE with IF condition>
default: N/A
Stress command for LWT performance test for mixed lwt load
default: N/A
Stress command for LWT performance test for mixed lwt load baseline
default: N/A
When define true, will install scylla cloud manager
default: N/A
When defined true, LDAP is going to be used.
default: N/A
When defined true, will create a docker container with LDAP and configure scylla.yaml to use it
default: N/A
When defined true, will create a docker container with LDAP and configure scylla.yaml to use it
default: N/A
When defined true, will install and start saslauthd service
default: N/A
This option indicates which server is going to be used for LDAP operations. [openldap, ms_ad]
default: N/A
When define true, will install scylla management
default: True
When defined true, will run node operations in parallel. Supported operations: startup
default: N/A
Port to be used by the manager to contact Prometheus
default: 5090
Url to the repo of scylla manager version used to upgrade the manager server
default: N/A
Url to the repo of scylla manager version used to upgrade the manager agents
default: N/A
A local directory of rpms to install a custom version on top of
the scylla installed (or from repo or from ami)
default: N/A
The port of scylla management
default: branch-4.8
Db type to install into db nodes, scylla/cassandra
default: scylla
the prefix of the name of the cloud instances, defaults to username
default: N/A
version name to report stats to Elasticsearch and tagged on cloud instances
default: N/A
Override the default hostname address of the sct test runner,
for the monitoring of the Nemesis.
can only work out of the box in AWS
default: N/A
Override the default hostname address of the sct test runner,
using ngrok server, see readme for more instructions
default: N/A
If True, all backtraces found in db nodes would be decoded automatically
default: True
Scylla will print kernel callstack to logs if True, otherwise, it will try and may print a message
that it failed to.
default: True
instance_provision: spot|on_demand|spot_fleet
default: spot
instance_provision_fallback_on_demand: create instance on_demand provision type if instance with selected 'instance_provision' type creation failed. Expected values: true|false (default - false
default: N/A
If reuse_cluster is set it should hold test_id of the cluster that will be reused.reuse_cluster: 7dc6db84-eb01-4b61-a946-b5c72e0f6d71
default: N/A
test id to filter by
default: N/A
How to select number of shards of Scylla. Expected values: default/random.
Default value: 'default'.
In case of random option - Scylla will start with different (random) shards on every node of the cluster
default: default
How to select the seeds. Expected values: random/first/all
default: all
Number of seeds to select
default: 1
list of email of send the performance regression test to
default: ['[email protected]']
Email subject postfix
default: N/A
Turn on sct profiling
default: N/A
Set type of ssh library to use. Could be 'fabric' (default) or 'libssh2'
default: libssh2
unlock specified experimental features
default: N/A
when enable scylla will use encryption on the server side
default: N/A
when enable scylla will use encryption on the client side
default: N/A
when enabled scylla will enforce mutual authentication when client-to-node encryption is enabled
default: N/A
when enabled scylla will enforce mutual authentication when node-to-node encryption is enabled
default: N/A
when enable or disable scylla hinted handoff (enabled/disabled)
default: disabled
which authenticator scylla will use AllowAllAuthenticator/PasswordAuthenticator
default: N/A
the username if PasswordAuthenticator is used
default: N/A
the password if PasswordAuthenticator is used
default: N/A
which authorizer scylla will use AllowAllAuthorizer/CassandraAuthorizer
default: N/A
run SLA nemeses if the test is SLA only
default: N/A
List if service level shares - how many server levels to create and test. Uses in SLA test.list of int, like: [100, 200]
default: [1000]
Port to configure for alternator in scylla.yaml
default: N/A
Type of dynamodb table to create with range key or not, can be:
HASH,HASH_AND_RANGE
default: HASH
Set the write isolation for the alternator table, see https://github.com/scylladb/scylla/blob/master/docs/alternator/alternator.md#write-isolation-policies for more details
default: N/A
If true, spawn a docker with a dns server for the ycsb loader to point to
default: N/A
If true, enable the authorization check in dynamodb api (alternator)
default: N/A
the aws_access_key_id that would be used for alternator
default: N/A
the aws_secret_access_key that would be used for alternator
default: N/A
When in multi region mode, run stress on loader that is located in the same region as db node
default: N/A
More arguments to append to scylla command line
default: --blocked-reactor-notify-ms 25 --abort-on-lsa-bad-alloc 1 --abort-on-seastar-bad-alloc --abort-on-internal-error 1 --abort-on-ebadf 1 --enable-sstable-key-validation 1
More arguments to append to oracle command line
default: --enable-cache false
More configuration to append to /etc/scylla/scylla.yaml
default: N/A
More arguments to append to scylla-node-exporter command line
default: N/A
Nemesis class to use (possible types in sdcm.nemesis).
Next syntax supporting:
- nemesis_class_name: "NemesisName" Run one nemesis in single thread
- nemesis_class_name: ":" Run in
parallel threads on different nodes. Ex.: "ChaosMonkey:2"
- nemesis_class_name: ": :" Run
in parallel threads and in
parallel threads. Ex.: "DisruptiveMonkey:1 NonDisruptiveMonkey:2"
default: NoOpMonkey
Nemesis sleep interval to use if None provided specifically in the test
default: 5
Sleep interval between nemesis operations for use in unique_sequence nemesis kind of tests
default: N/A
Run nemesis during prepare stage of the test
default: True
A seed number in order to repeat nemesis sequence as part of SisyphusMonkey
default: N/A
Add/remove nodes during GrowShrinkCluster nemesis
default: 1
Instance type to use for adding/removing nodes during GrowShrinkCluster nemesis
default: N/A
Used for scale test: max size of the cluster
default: N/A
Space node threshold before starting nemesis (bytes)
The default value is 6GB (6x1024^3 bytes)
This value is supposed to reproduce
scylladb/scylladb#1140
default: N/A
If true runs the nemesis only on non seed nodes
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
Url of the schema/configuration the gemini tool would use
default: N/A
gemini command to run (for now used only in GeminiTest)
default: N/A
Seed number for gemini command
default: N/A
table options for created table. example:
["cdc={'enabled': true}"]
["cdc={'enabled': true}", "compaction={'class': 'IncrementalCompactionStrategy'}"]
default: N/A
AWS image type of the loader node
default: N/A
AWS image type of the monitor node
default: N/A
AWS image type of the db node
default: N/A
AWS image type of the oracle node
default: N/A
instance type of the sct-runner node
default: N/A
AWS regions to use
default: N/A
AWS security groups ids to use
default: N/A
if true, create 'cluster' placement group for test case for low-latency network performance achievement
default: N/A
AWS subnet ids to use
default: N/A
AMS AMI id to use for scylla db node
default: N/A
AMS AMI id to use for loader node
default: N/A
AMS AMI id to use for monitor node
default: N/A
AMS AMI id to use for cassandra node
default: N/A
AMS AMI id to use for oracle node
default: N/A
default: N/A
default: N/A
default: N/A
root disk size in Gb for sct-runner
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
The max percentage of the on demand price we set for spot/fleet instances
default: N/A
if true, create extra network interface on each node
default: N/A
This is the name of the instance profile to set on all db instances
default: N/A
This is the name of the instance profile to set on all loader instances
default: N/A
the backend to be used for backup (e.g., 's3', 'gcs' or 'azure')
default: N/A
the bucket name to be used for backup (e.g., 'manager-backup-tests')
default: N/A
the AWS region of a bucket to be used for backup (e.g., 'eu-west-1')
default: N/A
If True, we use prepared VMs for loader (instead of using docker images)
default: N/A
list of files that should upload to /etc/scylla.d/ directory to override scylla config files
default: N/A
gcp project name to use
default: N/A
Supported: us-east1 - means that the zone will be selected automatically or you can mention the zone explicitly, for example: us-east1-b
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
If True, SCT configures a hybrid RAID of NVMEs and an SSD for scylla's data
default: N/A
default: N/A
default: N/A
Supported: eastus
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
K8S release channel name to be used. Expected values are: 'rapid', 'regular', 'stable' and '' (static / No channel).
default: N/A
Docker image to be used by Scylla operator to tune K8S nodes for performance. Used when k8s_enable_performance_tuning' is defined to 'True'. If not set then the default from operator will be used.
default: N/A
Define whether performance tuning must run or not.
default: N/A
default: N/A
Defines the type of the K8S local volume provisioner to be deployed. It may be either 'static' or 'dynamic'. Details about 'dynamic': 'dynamic': https://github.com/scylladb/k8s-local-volume-provisioner; 'static': sdcm/k8s_configs/static-local-volume-provisioner.yaml
default: N/A
Docker image to be used for installation of scylla operator.
default: N/A
Docker image to be used for upgrade of scylla operator.
default: N/A
Link to the Helm repository where to get 'scylla-operator' charts from.
default: N/A
Link to the Helm repository where to get 'scylla-operator' charts for upgrade.
default: N/A
Version of 'scylla-operator' Helm chart to use. If not set then latest one will be used.
default: N/A
Version of 'scylla-operator' Helm chart to use for upgrade.
default: N/A
Defines whether dataset uses for pre-fill cluster in functional test. Defined in sdcm.utils.sstable.load_inventory. Expected values: BIG_SSTABLE_MULTI_COLUMNS_DATA, MULTI_COLUMNS_DATA
default: N/A
The CPU limit that will be set for each Scylla cluster deployed in K8S. If not set, then will be autocalculated. Example: '500m' or '2'
default: N/A
The memory limit that will be set for each Scylla cluster deployed in K8S. If not set, then will be autocalculated. Example: '16384Mi'
default: N/A
default: N/A
Number of loader pods per loader cluster.
default: 3
default: N/A
default: N/A
default: N/A
Number of loader pods per loader cluster.
default: N/A
Defines how the loader pods must run. It may be either 'static' (default, run stress command on the constantly existing idle pod having reserved resources, perf-oriented) or 'dynamic' (run stress commad in a separate pod as main thread and get logs in a searate retryable API call not having resource reservations).
default: dynamic
Instance type for the nodes of the K8S auxiliary/default node pool.
default: N/A
Instance type for the nodes of the K8S monitoring node pool.
default: N/A
default: N/A
default: N/A
default: 10Gi
Defines whether the K8S API server logging must be enabled and it's logs gathered. Be aware that it may be really huge set of data.
default: N/A
Number of Scylla clusters to create in the K8S cluster.
default: 1
Defines whether we enable the scylla operator TLS feature or not.
default: N/A
Defines whether we install SNI and use it or not (serverless feature).
default: N/A
Defines whether we enable the alternator feature using scylla-operator or not.
default: N/A
Serverless configuration bundle file
default: N/A
Defines the type of the K8S 'Service' objects type used for ScyllaDB pods. Empty value means 'do not set and allow scylla-operator to choose'.
default: N/A
Defines the source of the IP address to be used for the 'broadcast_address' config option in the 'scylla.yaml' files. Empty value means 'do not set and allow scylla-operator to choose'.
default: N/A
Defines the source of the IP address to be used for the 'broadcast_rpc_address' config option in the 'scylla.yaml' files. Empty value means 'do not set and allow scylla-operator to choose'.
default: N/A
enables chaos-mesh for k8s testing
default: N/A
Number of nodes in auxiliary pool
default: N/A
Number of nodes in monitoring pool that will be used for scylla-operator's deployed monitoring pods.
default: N/A
Scylla manager docker image, i.e. 'scylladb/scylla-manager:2.2.1'
default: scylladb/scylla-manager:3.4.0
Scylla docker image repo, i.e. 'scylladb/scylla', if omitted is calculated from scylla_version
default: N/A
local docker network to use, if there's need to have db cluster connect to other services running in docker
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: 1000000
default: 1000
default: 1
Number of cassandra-stress processes
default: 1
Number of cassandra-stress processes for write workload
default: 1
Number of cassandra-stress processes for read workload
default: 1
Number of cassandra-stress processes for mixed workload
default: 1
default: N/A
Runs a background thread that issues reversed-queries on a table random partition by an interval
default: N/A
Runs a background thread that verifies Tombstones GC on a table by an interval
default: N/A
default: 1
default: N/A
default: 1
default: N/A
Command to create keysapce to be pre-create before running workload
default: N/A
CQL Commands to run after prepare stage finished (relevant only to longevity_test.py)
default: N/A
At the end of prepare stage, run major compaction and wait for this time (in minutes) for compaction to finish. (relevant only to longevity_test.py), Should be use only for when facing issue like compaction is affect the test or load
default: N/A
Choose a specific compaction strategy to pre-create schema with.
default: SizeTieredCompactionStrategy
Configure sstable size for the usage of pre-create-schema mode
default: N/A
When true, start cluster health checker for all nodes
default: True
A group of sub-parameters: validate_partitions, table_name, primary_key_column,
partition_range_with_data_validation, max_partitions_in_test_table.
1. validate_partitions - when true, validating the same number of rows-per-partition before/after a Nemesis.
2. table_name - table name to check for the validate_partitions check.
3. primary_key_column - primary key of the table to check for the validate_partitions check
4. partition_range_with_data_validation - Relevant for scylla-bench. A range (min - max) of PK values
for partitions to be validated by reads and not to be deleted during test. Example: 0-250.
5. max_partitions_in_test_table - Relevant for scylla-bench. Max partition keys (partition-count)
in the scylla_bench.test table.
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
number of tables to create for template user c-s
default: 1
Url to the repo of scylla manager version to upgrade to for management tests
default: N/A
Manager restore operation extra parameters: batch-size, parallel, etc.For example, --batch-size 2 --parallel 1. Provided string appends the restore cmd
default: N/A
Manager agent backup general configuration: checkers, transfers, low_level_retries. For example, {'checkers': 100, 'transfers': 2, 'low_level_retries': 20}
default: N/A
Name of backup snapshot to use in Manager restore benchmark test, for example, 500gb_2t_ics. The name provides the info about dataset size (500gb), tables number (2) and compaction (ICS)
default: N/A
Skip post-restore c-s verification read in the Manager restore benchmark tests
default: N/A
Nodetool refresh extra options like --load-and-stream or --primary-replica-only
default: N/A
Size of backup snapshot in Gb to be prepared to be prepared for backup
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands for warm-up before read workload.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
default: N/A
jobs to compare performance results with, for example if running in staging, we still can compare with official jobs
default: N/A
cassandra-stress user-profiles list. Executed in test step
default: N/A
cassandra-stress user-profiles list. Executed in prepare step
default: N/A
default: 50m
enable debug for cassandra-stress
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
Threads amount of c-s load for gradual performance test per sub-test. Example: {'read': 100, 'write': 200, 'mixed': 300}
default: N/A
Used for gradual performance test. Define throttle for load step in ops. Example: {'read': ['100000', '150000'], 'mixed': ['300']}
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
default: N/A
Assign new upgrade version, use it to upgrade to specific minor release. eg: 3.0.1
default: N/A
Assign target upgrade version, use for decide if the truncate entries test should be run. This test should be performed in case the target upgrade version >= 3.1
default: N/A
As for now, raft will be enable by default in all [upgrade] tests, so this flag will allow usto still run [upgrade] test without raft enabled (or disabling raft), so we will have bettercoverage
default: True
By default, the tablets feature is disabled. With this parameter, created for the upgrade test,the tablets feature will only be enabled after the upgrade
default: N/A
default: N/A
Upgrade system packages on nodes before upgrading Scylla. Enabled by default
default: True
default: N/A
Enable an option for installed 3.1.0 for work around a scylla issue if it's true
default: N/A
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
cassandra-stress commands.
You can specify everything but the -node parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: N/A
options will be used for enable encryption at-rest for tables
default: N/A
The time interval in minutes which gets waited before the KMS key rotation happens. Applied when the AWS KMS service is configured to be used.
default: N/A
An escape hatch to disable KMS for enterprise run, when needed, we enable kms by default since if we use scylla 2023.1.3 and up
default: N/A
How to transport logs: syslog-ng, ssh or docker
default: syslog-ng
Collect logs from instances and sct runner
default: N/A
Run post behavior actions in sct teardown step
default: N/A
Failure/post test behavior, i.e. what to do with the db cloud instances at the end of the test.
'destroy' - Destroy instances and credentials (default)
'keep' - Keep instances running and leave credentials alone
'keep-on-failure' - Keep instances if testrun failed
default: keep-on-failure
Failure/post test behavior, i.e. what to do with the loader cloud instances at the end of the test.
'destroy' - Destroy instances and credentials (default)
'keep' - Keep instances running and leave credentials alone
'keep-on-failure' - Keep instances if testrun failed
default: destroy
Failure/post test behavior, i.e. what to do with the monitor cloud instances at the end of the test.
'destroy' - Destroy instances and credentials (default)
'keep' - Keep instances running and leave credentials alone
'keep-on-failure' - Keep instances if testrun failed
default: keep-on-failure
Failure/post test behavior, i.e. what to do with the k8s cluster at the end of the test.
'destroy' - Destroy k8s cluster and credentials (default)
'keep' - Keep k8s cluster running and leave credentials alone
'keep-on-failure' - Keep k8s cluster if testrun failed
default: keep-on-failure
scylla option: internode_compression
default: N/A
scylla sub option of server_encryption_options: internode_encryption
default: all
The total size of the memory allocated to JMX. Values in MB, so for 1GB enter 1024(MB)
default: N/A
A flag that indicates whether or not to gather the prometheus stats at the end of the run.
Intended to be used in performance testing
default: N/A
More arguments to append to scylla_setup command line
default: N/A
Don't install/update ScyllaDB on DB nodes
default: N/A
Force running iotune on the DB nodes, regdless if image has predefined values
default: N/A
cdc-stressor command to read cdc_log table.
You can specify everything but the -node , -keyspace, -table, parameter, which is going to
be provided by the test suite infrastructure.
multiple commands can passed as a list
default: cdc-stressor -stream-query-round-duration 30s
Add cdclog reader stats to ES for future performance result calculating
default: N/A
If set to True the test will be stopped immediately when stress command failed.
When set to False the test will continue to run even when there are errors in the
stress process
default: True
retrieving data from multiple streams in one poll
default: True
Use legacy cluster initialization with autobootsrap disabled and parallel node setup
default: N/A
Availability zone to use. Specify multiple (comma separated) to deploy resources to multi az (works on AWS).
"Same for multi-region scenario.
default: N/A
Try all availability zones one by one in order to maximize the chances of getting
the requested instance capacity.
default: N/A
Number of nodes to upgrade and rollback in test_generic_cluster_upgrade
default: N/A
Whether to upgrade sstables as part of upgrade_node or not
default: N/A
Stress command to be run before upgrade (preapre stage)
default: N/A
Stress command to be run during the upgrade - user should take care for suitable duration
default: N/A
Stress command to be run after full upgrade - usually used to read the dataset for verification
default: N/A
Link to the git repository with Jepsen Scylla tests
default: https://github.com/jepsen-io/scylla.git
Jepsen test command (e.g., 'test-all')
default: ['test-all -w cas-register --concurrency 10n', 'test-all -w counter --concurrency 10n', 'test-all -w cmap --concurrency 10n', 'test-all -w cset --concurrency 10n', 'test-all -w write-isolation --concurrency 10n', 'test-all -w list-append --concurrency 10n', 'test-all -w wr-register --concurrency 10n']
possible number of reruns of single Jepsen test command
default: 1
Jepsen test run policy (i.e., what we want to consider as passed for a single test)
'most' - most test runs are passed
'any' - one pass is enough
'all' - all test runs should pass
default: all
Limit severity level for event types
default: N/A
Configure rsyslog on Scylla nodes to send logs to monitoring nodes
default: N/A
Limit number events in email reports
default: 10
Number of additional data volumes attached to instances
if data_volume_disk_num > 0, then data volumes (ebs on aws) will be
used for scylla data directory
default: N/A
Type of addtitional volumes: gp2|gp3|io2|io3
default: N/A
Size of additional volume in GB
default: N/A
Number of iops for ebs type io2|io3|gp3
default: N/A
Throughput in MiB/sec for ebs type gp3. Min is 125. Max is 1000.
default: N/A
Flag for running db node benchmarks before the tests
default: N/A
nemesis_selector gets a list of "nemesis properties" and filters IN all the nemesis that has
ALL the properties in that list which are set to true (the intersection of all properties).
(In other words filters out all nemesis that doesn't ONE of these properties set to true)
IMPORTANT: If a property doesn't exist, ALL the nemesis will be included.
default: N/A
nemesis_exclude_disabled determines whether 'disabled' nemeses are filtered out from list
or are allowed to be used. This allows to easily disable too 'risky' or 'extreme' nemeses by default,
for all longevities. For example: it is unwanted to run the ToggleGcModeMonkey in standard longevities
that runs a stress with data validation.
default: True
Multiply the list of nemesis to execute by the specified factor
default: 6
nemesis_double_load_during_grow_shrink_duration / SCT_NEMESIS_DOUBLE_LOAD_DURING_GROW_SHRINK_DURATION
After growing (and before shrink) in GrowShrinkCluster nemesis it will double the load for provided duration.
default: N/A
Number of of raid level: 0 - RAID0, 5 - RAID5
default: N/A
Don't install anything but node_exporter to the loaders during cluster setup
default: N/A
Dict of the images to use for the stress tools
default: N/A
Configure Scylla networking with single or multiple NIC/IP combinations.
It must be defined for listen_address and rpc_address. For each address mandatory parameters are:
- address: listen_address/rpc_address/broadcast_rpc_address/broadcast_address/test_communication
- ip_type: ipv4 or ipv6
- public: false or true
- nic: number of NIC. 0, 1
Supported for AWS only meanwhile
default: N/A
Control reporting to argus
default: True
set c-s parameter '-pop' with gauss/uniform distribution for
performance gradual throughtput grow tests
default: N/A
Optional. Allows to pass through custom rune script parameters to the 'latte schema' command.
default: N/A
Number of loaders which should be added per step
default: N/A
Number of threads for c-s command
default: N/A
Number of threads which should be added on per step
default: N/A
Duration of time for stress round
default: 15m
Max relative difference between best and current throughput,
if current throughput larger then best on max_rel_diff, it become new best one
default: N/A
Number of stress processes per loader
default: N/A
add/remove num of process on each round
default: N/A
Enable hdr histogram logging for cs
default: N/A
Stop sct performance test if hardware performance test failed
Hardware performance tests runs on each node with sysbench and cassandra-fio tools.
Results stored in ES. HW perf tests run during cluster setups and not affect
SCT Performance tests. Results calculated as average among all results for certain
instance type or among all nodes during single run.
if results for a single node is not in margin 0.01 of
average result for all nodes, hw test considered as Failed.
If stop_on_hw_perf_failure is True, then sct performance test will be terminated
after hw perf tests detect node with hw results not in margin with average
If stop_on_hw_perf_failure is False, then sct performance test will be run
even after hw perf tests detect node with hw results not in margin with average
default: N/A
Use custom ES index for storing test results
default: N/A
Defines how many regions must be simulated on the Scylla config side. If set then
nodes will be provisioned only using the very first real region defined in the configuration.
default: N/A
Forces GossipingPropertyFileSnitch (regardless endpoint_snitch) to simulate racks.
Provide number of racks to simulate.
default: N/A
Use dns names instead of ip addresses for nodes in cluster
default: N/A
Enable validation for large cells in system table and logs
default: N/A
Run commit log check thread if commitlog_use_hard_size_limit is True
default: True
Configuration for additional validations executed after the test
default: {'scrub': {'enabled': False, 'timeout': 1200, 'keyspace': '', 'table': ''}, 'test_error_events': {'enabled': False, 'failing_events': [{'event_class': 'DatabaseLogEvent', 'event_type': 'RUNTIME_ERROR', 'regex': '.runtime_error.'}, {'event_class': 'CoreDumpEvent'}]}}
reserves instances capacity for whole duration of the test run (AWS only).
Fallbacks to next availabilit zone if capacity is not available
default: N/A
Scylla build date from which bisecting should start.
Setting this date enables bisection. Format: YYYY-MM-DD
default: N/A
Scylla build date until which bisecting should run. Format: YYYY-MM-DD
default: N/A
Enable validation for large cells in system table and logs
default: N/A
configuration for setup up kafka connectors
default: N/A
Run scylla-doctor in artifact tests
default: N/A
Skip selected stages of a test scenario
default: N/A
If True, enable support in sct of zero nodes(configuration, nemesis)
default: N/A
Number of zero token nodes in cluster. Value should be set as "0 1 1"
for multidc configuration in same manner as 'n_db_nodes' and should be equal
number of regions
default: N/A
Instance type for zero token node
default: i4i.large
AWS account id on behalf of which the test is run
default: N/A