| title | pg hardstorage repo init | ||
|---|---|---|---|
| description | CLI reference for the pg hardstorage repo init command. | ||
| tags |
|
Create a new repository at the given URL
Initialise a new pg_hardstorage repository at the given URL.
Supported schemes: file://, s3:///
The operation is atomic and race-safe: if two processes init the same URL concurrently, exactly one wins and the other receives a conflict error.
WORM (write-once-read-many):
--worm-mode {compliance|governance} \
--worm-retention <duration> # 7y, 30d, 8760h
Records a retention policy in HSREPO that propagates to every committed object's PUT (chunks, manifests, replicas, audit events). Compliance mode is the regulatory-grade posture (even root credentials cannot delete before the deadline). Governance mode allows IAM principals with the BypassGovernance permission to delete. WORM is set at init time only — flipping it on later would produce a mixed-fleet situation operators can't reason about.
Retention units: y (365-day years), d (days), h (hours), m (minutes).
Compression:
--compression {fast|balanced|max}
Picks the zstd encoder level for new chunks. Profiling under a write-heavy workload (10 GB pgbench seed + sustained UPDATE load) showed the v0.1..default ("balanced", ~zstd level 7) burned ~40% of pg_hardstorage CPU.
fast ~zstd level 3. Halves zstd CPU; ~10-15% larger on disk. Recommended for write-heavy clusters where wal-stream CPU matters more than disk bytes. balanced ~zstd level 7. v0.1..default. Sweet spot for the median operator. max ~zstd level 11. 2-3x more CPU than balanced; ~5% smaller on disk. Archive-tier backups read rarely.
Set at init time and not changed after. A repo holding a mix of levels still reads back fine — the decoder handles every level — but the operator's "what does my CPU/disk trade-off look like?" answer is more legible when the level is stable across the repo.
pg_hardstorage repo init <url> [flags]
--compression string zstd encoder level: fast | balanced | max (default: balanced)
-h, --help help for init
--worm-mode string WORM retention mode: compliance | governance (set with --worm-retention)
--worm-retention string WORM retention duration: e.g. 7y, 30d, 8760h (required with --worm-mode)
--airgapped airgapped: strict refuse outbound endpoints (LLM providers, sinks, OTLP collectors) outside loopback / RFC1918 / explicit airgap.allowlist. Also enabled by PG_HARDSTORAGE_AIRGAPPED=1 or airgapped: strict in the config file.
-c, --config string path to config file (default: XDG/FHS lookup)
--cpu-profile go tool pprof <path> write a pprof CPU profile to this path for the duration of the command (go tool pprof <path> to analyse). Off when empty.
--mem-profile string write a pprof heap profile to this path at command exit. Off when empty.
--no-color disable ANSI color in text output
--on-error-llm on a structured-error failure, drop into the matching LLM helper skill (auto_on_error trigger). Also enabled by PG_HARDSTORAGE_ON_ERROR_LLM=1.
--otel-endpoint string OpenTelemetry OTLP/HTTP endpoint (e.g. http://otel-collector:4318); empty disables tracing
--otel-stdout also export OpenTelemetry traces to stderr (useful for dev)
-o, --output string output format: text|json|ndjson|yaml|template|csv|markdown|html|tap|junit|pdf (default: text on TTY, json off-TTY)
--profile-port go tool pprof http://127.0.0.1:6060/debug/pprof/profile?seconds=30 if non-zero, expose net/http/pprof on 127.0.0.1:<port> for live profiling of long-running commands (e.g. go tool pprof http://127.0.0.1:6060/debug/pprof/profile?seconds=30). Off when zero.
-q, --quiet suppress non-essential output
--template string Go text/template applied when --output template (or implied if --template is set without --output)
- pg_hardstorage repo - Manage the repository