Collector Readiness Check

Before adding a new collector to an auto-balance group, verify it can reach all devices in that group. Once the collector joins, LM starts assigning devices to it automatically — devices it can’t reach will generate monitoring errors.

The standard workflow:

Build the new collector and leave it outside any group.
Discover devices and protocols using tools/elm-collector-readiness.sh.
Generate the test script (stdout redirect or clipboard).
Test reachability from the new collector in LM Collector Debug.
Fix any failures before moving the collector into the group.

See also:

collectors.md for health checks and auto-balance group queries

Step 1 — Find the auto-balance group
Step 2 — Discover devices and generate the test script
Step 3 — Inject credentials (optional)
Step 4 — Run the test from the new collector
Interpreting results
- SNMP TIMEOUT
- WMI (tcp-135)
Mode C — LM API directly, no elm
meta

Step 1 — Find the auto-balance group

Run with no arguments to list auto-balance groups:

tools/elm-collector-readiness.sh

Auto-balance groups:

id    name                      collectors  threshold
----  ------------------------  ----------  ---------
42    My Region Collectors      3           500
87    APAC Collectors           2           500

Note the ID of the group you are adding the new collector to.

Step 2 — Discover devices and generate the test script

The script always renders the Groovy test script to stdout. Redirect it to a file or pipe to clipboard.

By group ID — write to file:

tools/elm-collector-readiness.sh --id 42 > /tmp/check.groovy

By group name — write to file:

tools/elm-collector-readiness.sh --name "My Region Collectors" > /tmp/check.groovy

Copy directly to clipboard (macOS):

tools/elm-collector-readiness.sh --id 42 | pbcopy

With a non-default elm profile:

tools/elm-collector-readiness.sh --id 42 --profile prod > /tmp/check.groovy

--profile defaults to config (the same default as elm — reads config.ini).

Status messages go to stderr so they don’t pollute the redirected output:

Group:       My Region Collectors (id=42)
Collectors:  3
AutoBalance: true

Fetching devices in group 42...
Devices found: 47

Device                           IP/Hostname             Protocols
-------------------------------- ----------------------  ---------
server01                         10.0.1.10               ping, snmp, tcp-22
windows-box                      10.0.1.20               ping, tcp-135
api-device                       10.0.1.30               ping, tcp-80, tcp-443

Protocol detection uses autoProperties set by LM Active Discovery on each device. The IP/hostname used is the name field — the address LM uses to reach the device, not displayName.

autoProperty	Value	Test added
`auto.snmp.operational`	`true`	SNMP probe (UDP 161)
`auto.network.listening_tcp_ports`	contains `22`	TCP port 22 (SSH)
`auto.network.listening_tcp_ports`	contains `80`	TCP port 80 (HTTP)
`auto.network.listening_tcp_ports`	contains `135`	TCP port 135 (WMI/RPC)
`auto.network.listening_tcp_ports`	contains `443`	TCP port 443 (HTTPS)
(always)	—	Ping (ICMP)

If a device has no Active Discovery data yet (no auto.network.listening_tcp_ports), only ping is tested. Run Active Discovery on the group in LM before using this tool for best results.

Step 3 — Inject credentials (optional)

By default the rendered script has blank credential fields; you fill them in manually in the LM debug console before running. If you want credentials pre-filled from elm’s config, add --creds:

# Default elm profile (config.ini)
tools/elm-collector-readiness.sh --id 42 --creds | pbcopy

# Non-default profile
tools/elm-collector-readiness.sh --id 42 --creds --profile prod > /tmp/check.groovy

Note: the rendered script from --creds contains your LM API credentials. Do not commit it to git or share it.

With credentials injected, the script can also call the LM API directly from the collector to re-discover the device list — useful if you want to verify without trusting the workstation-side output.

Step 4 — Run the test from the new collector

Open the LM portal and navigate to the new collector’s device.
Go to Collector Debug → Script tab.
Paste the rendered Groovy (from file or clipboard) and run it.

Example output:

47 devices (pre-filled by elm)

Device                           IP/Hostname          ping      snmp      tcp-22    tcp-135
-------------------------------- -------------------- --------- --------- --------- ---------
server01                         10.0.1.10            PASS      PASS      PASS      -
windows-box                      10.0.1.20            PASS      -         -         PASS
api-device                       10.0.1.30            PASS      -         -         -
unreachable-host                 10.0.2.99            FAIL      -         FAIL      -

FAILURES — investigate before adding this collector to the group:
  - unreachable-host  ping
  - unreachable-host  tcp-22

Interpreting results

Result	Meaning
`PASS`	Connection succeeded
`FAIL`	Connection refused or timed out — routing or firewall issue
`TIMEOUT`	SNMP only: no UDP response within timeout
`-`	Protocol not expected for this device; skipped

SNMP TIMEOUT

TIMEOUT on SNMP does not necessarily mean the device is unreachable. SNMP agents that enforce community strings silently drop probes with unknown communities instead of sending an error response. The probe uses community public; if the device uses a different community, you will see TIMEOUT even though the agent is running and the port is open.

If ping passes but snmp shows TIMEOUT, check the device’s snmp.community property in LM and verify the collector can reach UDP 161 from the network level.

WMI (tcp-135)

TCP 135 is the WMI/DCOM endpoint mapper. A passing TCP-135 check means the Windows RPC endpoint is reachable from the new collector, which is the necessary precondition for WMI collection. It does not test WMI credentials.

Mode C — LM API directly, no elm

If elm is not available, open tools/lm-collector-reachability-check.groovy.j2 in an editor, clear the `` placeholder (leave []), and fill in the credentials block:

def ACCESS_ID    = "your-access-id"   // remove after use
def ACCESS_KEY   = "your-access-key"  // remove after use
def ACCOUNT_NAME = "acme"
def GROUP_ID     = 42
// DEVICES_JSON stays as []

The script calls the LM API directly to discover devices. Remove the credentials from the script after use — the LM debug console history may retain them.