Usage

Run Refresh Jobs

Trigger refresh playbooks across your nodes:

dmtri refresh --network HL --station ATH --starttime 2024-01-01

Track Job Status

Monitor the status of asynchronous jobs:

dmtri track

Fix Inconsistencies

Automatically repair the inconsistencies found in an eida-consistency report. A report lists streams where the availability view and dataselect disagree; dmtri fix reads that report and, for each affected stream, re-runs the WFCatalog collector and rebuilds the availability view for the exact time window.

Point it at a report file or URL:

dmtri fix https://eida-oculus.orfeus-eu.org/consistency/NOA/2026/NOA_2026-06-07_140510.json

Fix only specific entries from the report (by their index):

dmtri fix report.json --index 2 --index 16

After a fix, the availability service keeps cached answers for ~20 minutes, so a stream may still look unfixed for a short while. Re-check once the cache expires — this only verifies, it changes nothing:

dmtri fix --verify-only report.json

Prerequisite: the eida-consistency CLI

dmtri fix reads the report through eida-consistency (version 0.5.1 or newer, for its explore --json output). If it isn't already on the machine, install it any one of these ways:

uv tool install eida-consistency     # recommended — puts it on your PATH
pipx install eida-consistency
pip install eida-consistency

You don't strictly have to install it: if uv is present, dmtri fix will run it on demand via uvx eida-consistency. And if it's installed somewhere off your PATH, point dmtri at it:

export DMTRI_EIDA_CONSISTENCY='/full/path/to/eida-consistency'

If it's missing (or too old), dmtri fix stops before touching anything and prints the exact install/upgrade command to run.

Fix Missing PSDs

Sometimes a day has data but no PSD. eida-consistency finds these days. dmtri fix --psd makes the PSDs again.

You need eida-consistency 0.7.1 or newer.

1. Find the problems

eida-consistency consistency --node NOA

This writes a report, for example reports/noa_20261008.json.

2. Fix them

dmtri fix reports/noa_20261008.json --psd

dmtri shows what it will do and asks Proceed? (y/N). Type y.

3. Check

Wait 20 minutes, then:

eida-consistency rerun reports/noa_20261008.json

RESOLVED means fixed. PERSISTS means still broken. See Troubleshooting.

What dmtri does

For each day without a PSD, dmtri takes that day, plus 15 days before and 15 days after. It runs one seedpsd command on those days:

seedpsd-cli data mseed --update --list <files>

dmtri never fixes a whole station at once, because that can take days. If a station has many more broken days, dmtri tells you:

  repaired : 7 window(s)
  skipped  : 2 window(s) (nothing in this archive to compute from)
  failed   : 0 window(s)

ℹ These channels have more broken days than the window covered.
  They need a backfill, not a repair:
    HP.EFP..HHZ  2919 missing + 0 invalid day(s) remain

To fill those days, run seedpsd-cli data load on the seedpsd machine when you have time.

Options

Option	What it does
--psd	Fix availability first, then PSDs
--psd-only	Fix PSDs only
--psd-window 5	Use 5 days before and after, instead of 15
--required-only	Fix only days from 2024-01-01 on, when PSDs became required
--no-confirm	Start without asking

Operation Details

dmtri supports two main operations for both metadata and data: refresh and clean. These affect the following components:

Refresh

• Metadata
  • seedpsd: Scans for new metadata.
  • availability: Refreshes the view in MongoDB.
• Data
  • Identifies data files in the archive for the specified epoch (by NSLC and time range).
    • seedpsd: Recalculates seedPSD for new files.
    • wfcatalog: Runs the collector on new data files.

Clean

• Metadata
  • seedpsd: Removes the metadata entries.
  • availability: Refreshes the view in MongoDB.
• Data
  • Identifies data files in the archive for the specified epoch.
    • seedpsd: Removes associated seedPSD data (via CLI).
    • wfcatalog: Removes files using the --delete option in the collector.

---

Source: EIDA/dmtri (main)