Multinode

[exaSR]

This PR adds multinode cluster support for Exasol and ClickHouse, enabling benchmarks on distributed database configurations. It also introduces comprehensive configuration validation and a new benchkit check command for config inspection.

Multinode Cluster Support

• Simple configuration: Set node_count: N in system setup to deploy N-node clusters
• Automatic infrastructure: Terraform provisions multiple instances per system (exasol-node0, exasol-node1, etc.)
• SSH key distribution: Automatic inter-node SSH setup for cluster communication
• System-specific implementations:
  • Exasol: c4 cluster tool with space-separated IP lists
  • ClickHouse: Sharding without replication via XML config files (remote_servers.xml, macros.xml)

Config Validation (benchkit/config.py)

• System name validation: Must be valid bash variable name ([a-zA-Z_][a-zA-Z0-9_]*)
• IP variable validation: $SYSTEMNAME_PRIVATE_IP must reference configured systems
• Unique system names: No duplicates allowed
• Instance config matching: env.instances keys must match system names
• Method-specific fields: Validates required fields (e.g., Exasol installer requires c4_version, working_copy, etc.)
• Workload validation: Positive scale_factor, runs_per_query; valid workload names

New benchkit check Command

benchkit check -c config.yaml      # Validate and display config
benchkit check -c config.yaml -v   # Verbose mode with all setup params

• Displays organized sections with Rich panels
• Shows validation status with clear error messages
• IP variable validation indicators
• Password masking in verbose mode

Infrastructure Improvements

• Inter-node communication: Security group rules for TCP/UDP between cluster nodes
• EBS disk support: Added ebs as alias for gp3 with auto-tuned IOPS/throughput
• Storage flexibility: Support for local NVMe, EBS, and root filesystem storage

Other Improvements

• Multiuser benchmark support (num_streams, randomize, random_seed)
• Enhanced logging and error messages
• Fixed single-user handling edge cases
• Improved report generation for multinode benchmarks