Version: Current
Historical release snapshot. These docs are frozen. Read current docs.
Massprox Configuration
Massprox Configuration
Section titled “Massprox Configuration”Status as of 2026-11-05: Massprox is partially operational. The gatherer loop may be inactive. Verify live state before changes.
Massprox reads most settings from environment variables. Keep runtime operational settings auditable and DB-first where the service supports it. Use environment variables only for bootstrap and process-level limits.
Core variables
Section titled “Core variables”| Variable | Default | Description |
|---|---|---|
OUT |
proxies |
State directory |
TARGET |
1000000 |
Candidate target per run |
VERIFY_CONC |
2048 |
HTTP verify slots (capped) |
RECHECK_CONC |
1024 |
Recheck slots |
TCP_CONC |
8192 |
TCP dial slots (capped) |
SCRAPE_CONC |
128 |
Scrape slots (capped) |
TCP_TIMEOUT |
400ms |
Dial timeout |
State layout
Section titled “State layout”Massprox keeps its durable state in PostgreSQL (owner ruling 2026-11-09:
durable state lives only in PostgreSQL). Per definitions/massprox.md:
- Schemas
massproxandmassprox_opshold the frontier, operational state and audit rows. - Raw lists (active, candidate, merged,
sources/,ranked/) are files with a hash and a retention rule; they are inputs and outputs, not the state store. - The embedded-database store and its migration tool
(
massprox-migrate-sqlite) are migration inputs only and are cut.
Under OUT you may still see list files from an older layout (active.txt,
candidates.txt, merged.txt, sources/, ranked/) and a status snapshot.
Treat them as operational data: read the Postgres state, and do not commit any
file under OUT.
Safety and egress
Section titled “Safety and egress”- Keep egress guard checks enabled.
- Do not change routing to bypass the controlled gateway.
- Do not enable direct IPv6 egress for UID 988.
- Do not use settings to probe restricted addresses.
- Limit concurrency to avoid resource exhaustion.
See Verify Egress Compliance and Massprox Operator Runbook.
Operational notes
Section titled “Operational notes”- Check systemd unit and drop-ins before edits.
- Use bounded worker pools.
- Monitor latency, failure reasons, and verified count.
- Record config changes in the approved audit path.