runtime settings

Runtime Settings

You will learn which knobs control daemon behavior, how to tune concurrency, set health thresholds, and configure web publishing.

Where runtime settings live

All runtime settings go in runtime.yaml at the catalog root. Example from the shipped catalog:

runtime:
  base_dir: ${BASE_DIR-${HOME}/ipsets}
  processing_interval_minutes: 5
  min_run_interval_seconds: 30
  max_processing_workers: 2
  max_background_workers: 1
  web_artifact_cache_max_entries: 2048
  web_artifact_cache_max_bytes: 67108864
  web_artifact_cache_max_file_bytes: 8388608
  feed_health_single_observation_grace_minutes: 14400
  feed_health_default_healthy_cadence_minutes: 10080
  feed_health_default_risky_cadence_minutes: 43200
  feed_health_archival_threshold_minutes: 86400
  web_url: https://iplists.firehol.org/ipsets/
  public_base_url: https://iplists.firehol.org

Concurrency domains

The daemon separates work into four independent concurrency pools.

Domain	Setting	Default	Controls
Download workers	parallel_downloads	5	Upstream HTTP/HTTPS acquisition and merge composition
Feed-processing workers	max_processing_workers	2	Turning staged downloads into committed feed outputs
Heavy-phase workers	max_heavy_phase_workers	auto (min(CPU, 8))	Pairwise comparisons, GeoIP/ASN/bogon fan-out
Background workers	max_background_workers	1	Startup repair, health-transition refreshes, deferred maintenance

Background work is intentionally low-priority. It prefers finishing later over competing with the main pipeline for CPU and memory.

Processing cadence

• processing_interval_minutes — how often the processing queue drains automatically. Default: 5.
• min_run_interval_seconds — minimum time between scheduler runs. Prevents rapid re-scheduling. Default: 30.

Download and parsing limits

These settings control acquisition and normalization cost.

Setting	Default	Purpose
max_connect_time	10 seconds	Maximum time to establish an upstream connection
max_download_time	300 seconds	Maximum time allowed for one download
max_download_size	100 MiB when unset or 0	Maximum response body size; -1 disables the cap
ignore_repeating_download_errors	10	Accepted runtime field. Current retry timing is driven by failure count and health class, not this value.
parallel_dns_queries	10	DNS lookups allowed in parallel while parsing hostname-based input
user_agent	FireHOL default	User-Agent sent to upstream HTTP servers

Artifact parents can override max_download_size in their own artifact definition.

Cache limits for published artifacts

The public web server caches generated JSON and static artifacts in memory. Raw .ipset/.netset downloads are streamed separately and do not use this cache.

Setting	Default	Purpose
web_artifact_cache_max_entries	2048	Maximum number of cached files
web_artifact_cache_max_bytes	64 MiB	Total cache size across all entries
web_artifact_cache_max_file_bytes	8 MiB	Maximum single file size in cache

Output reduction and charts

Setting	Default	Purpose
ipset_reduce_factor	20	Reduction factor used when producing reduced ipset-compatible outputs
ipset_reduce_entries	65536	Target entry count for reduced outputs
web_charts_entries	500	Number of historical points used for generated chart data

Health thresholds

Health states determine whether a feed is considered healthy, delayed, risky, unmaintained, or archived.

Setting	Default	Purpose
feed_health_single_observation_grace_minutes	14400 (10 days)	Grace period before a feed with only one observation gets health-classified
feed_health_default_healthy_cadence_minutes	10080 (7 days)	Default upper bound for "healthy" age
feed_health_default_risky_cadence_minutes	43200 (30 days)	Default upper bound for "risky" age
feed_health_archival_threshold_minutes	86400 (60 days)	Continuous unavailable duration before archival

Category-specific overrides live in feed_health_category_thresholds. For example, intrusion feeds use tighter thresholds than special_use feeds because intrusion feeds update more frequently.

Web publishing

Setting	Purpose
public_base_url	Externally visible base URL of the public website. Used for admin-to-public navigation links.
web_url	Published feed-detail prefix used in generated metadata and output files. May include a path like /ipsets/.
local_copy_url	Base URL for raw file downloads.
web_dir	Local directory for published web files. Can be a separate git repository.
web_dir_for_ipsets	Local directory for downloadable .ipset and .netset files.
web_owner	Optional filesystem owner applied to published web files.
github_changes_url	URL prefix used in metadata links to feed-output commits.
github_setinfo	URL prefix used in metadata links to setinfo files.

public_base_url and web_url serve different purposes. public_base_url is the website root. web_url is the feed-detail path prefix. Do not use them interchangeably.

Git push options

Setting	Default	Purpose
push_to_git	false	Enable git commits/pushes for generated outputs
push_to_git_merged	true	Commit merged output after processing
push_to_git_commit_options	empty	Extra options passed to git commit
push_to_git_push_options	empty	Extra options passed to git push
push_to_git_web	false	Also commit the web_dir tree if it is a separate repository

Runtime paths

These settings control where the daemon stores state. The installed systemd unit normally sets matching environment variables instead of requiring YAML edits.

Setting	Purpose
base_dir	Committed .ipset, .netset, .source, .setinfo, and enable-marker files
config_file	Legacy bash config file path used by compatibility loaders
run_parent_dir	Parent directory for the process lock
lock_file	Explicit lock-file path
cache_dir	Scheduler/runtime cache directory
lib_dir	Binary snapshots, history ledgers, retention data, provider state, and entity sidecars
admin_supplied_ipsets	Admin-managed supplemental catalog directory
distribution_supplied_ipsets	Distribution-packaged supplemental catalog directory
user_supplied_ipsets	User-managed supplemental catalog directory
history_dir	Feed history snapshot directory
errors_dir	Download error-log directory
tmp_dir	Temporary download, extraction, and staging directory

See Filesystem Layout for the installed directory tree.

Network trust and local apply

Setting	Default	Purpose
trust_proxy_headers	false	Trust X-Forwarded-For and X-Real-IP for client IP detection
trust_cloudflare_headers	false	Trust CF-Connecting-IP for client IP detection
ipsets_apply	true for root, false for non-root	Apply generated sets to kernel ipset when supported

Only enable trusted-header settings when every request reaches the daemon through the trusted proxy. Direct client access with these enabled lets clients choose their apparent IP address.

Environment variable expansion

Runtime settings support ${VAR-default} shell-style expansion. For example:

base_dir: ${BASE_DIR-${HOME}/ipsets}

This resolves $BASE_DIR if set, otherwise falls back to $HOME/ipsets.