Request
The scan settings must eventually be configured from the web interface, not through environment variables:
Why this is more than a convenience
Environment variables are the wrong home for these three because changing them requires restarting the container, and the values they hold are exactly the ones an operator needs to tune iteratively, against a live network, while watching what changes:
- The timeout decides what the scan misses — a device that answers more slowly is recorded as absent, not as unknown. Finding the right value means trying one, looking at the result, and trying another.
- The concurrency trades politeness against wall-clock, and the acceptable point differs per network.
- Neither has a defensible universal default; both are properties of this network.
A restart per experiment makes that loop expensive enough that people stop tuning and accept whatever the default gives them — which, for a product whose thesis is the accuracy of what it observes, is a poor outcome.
Notes
Request
The scan settings must eventually be configured from the web interface, not through environment variables:
OPENCMDB_SCAN_CIDR— already tracked in Configure scan CIDRs from the web UI — multiple subnets of heterogeneous sizes (replace OPENCMDB_SCAN_CIDR) #3 / Manage scan subnets (multiple CIDRs) from the web UI #4OPENCMDB_SCAN_CONCURRENCY— probes in flightOPENCMDB_SCAN_TIMEOUT_MS— how long one probe waitsWhy this is more than a convenience
Environment variables are the wrong home for these three because changing them requires restarting the container, and the values they hold are exactly the ones an operator needs to tune iteratively, against a live network, while watching what changes:
A restart per experiment makes that loop expensive enough that people stop tuning and accept whatever the default gives them — which, for a product whose thesis is the accuracy of what it observes, is a poor outcome.
Notes