Fix incorrect configurationId when multiple DCRs share the same counters for ME metrics

[simathih]

When multiple DCRs configured the same performance counters, all metrics were sent to ME tagged with only the first DCR's configurationId (due to an early break). ME uses this tag to route data per-DCR, so all subsequent DCRs received incorrectly attributed metrics.

Changes

• parse_config in telegraf_config_handler.py
  • Collect all unique configurationIds across all fields in a plugin before emitting any config
  • For each configId, emit a separate telegraf config file named <omiclass>-<configId>.conf (previously one file per plugin, e.g. filesystem.conf)
  • Each generated file carries only its own configurationId in the [inputs.<plugin>.tags] block

Before (one file per plugin, wrong configId for DCR 2+):

filesystem.conf          # configurationId="dcr-fa9e2b03..." for all DCRs

After (one file per plugin per DCR, correct configId in each):

filesystem-dcr-c93b5596....conf   # configurationId="dcr-c93b5596..."
filesystem-dcr-fa9e2b03....conf   # configurationId="dcr-fa9e2b03..."

[simathih]

it will look like this now:

from:

[simathih]

@copilot please review this

[Copilot]

@copilot please review this

The fix looks correct for the AMA (non-LAD) multiple DCR case. Here's my review:

What the fix does:

• Old behavior: only the first configId from counterConfigIdMap was used per plugin (the old code had a break after getting the first configId), so all metrics for a plugin across multiple DCRs were tagged with the same (wrong) configurationId.
• New behavior: collects all unique configIds across all fields in a plugin (lines 174–179), then generates a separate config file per configId (e.g., filesystem-dcr-abc123.conf, filesystem-dcr-def456.conf), each with the correct configurationId tag. This matches the before/after screenshots above.

One observation:
Line 187 initializes rate_specific_aggregator_str but it's never used (was the same before this PR). Minor nit.

One concern:
At line 176, counterConfigIdMap is looked up using telegraf_json[omiclass][plugin][field]["displayName"]. For non-LAD (AMA), the displayName stored in telegraf_json is the full counter name (line 126), which matches the key used when building counterConfigIdMap (line 100–101). This is correct ✓.

For LAD (is_lad=True), the displayName stored in telegraf_json is counter.split("->")[1] (line 124 — just the short name), while counterConfigIdMap uses the full counter as key. This could cause a KeyError for LAD. However, since LAD is deprecated and the LAD data format from parse_lad_perf_settings doesn't include configurationId (which would fail earlier at line 105 anyway), this is a pre-existing concern rather than a new regression.

Summary: The fix is correct for the intended AMA use case and the file-per-DCR approach matches what's shown in the screenshots.

[AaronRM]

This change will generate an [[inputs.plugin]] config per DCR (potentially resulting in telegraf collecting duplicate metrics per collection interval if they are duplicated across DCRs). This appears to be unavoidable with Telegraf. Just calling it out for awareness.