Listing Jobs
# Show running processes
verdi process list
# Show all processes (including finished)
verdi process list -a
# Filter by process label (wildcards supported)
verdi process list -L %Base% -a -p 30
# Show only failed jobs
verdi process list -X
# Limit results
verdi process list -a -l 50Useful options:
-p, --past-days— only show jobs from last N days-L, --process-label— filter by label (supports%wildcards)-S, --process-state— filter by state:created,running,waiting,finished,excepted,killed-X, --failed— show only failed jobs-G, --group— only jobs in a specific group
Troubleshooting
Get detailed report for a specific job:
verdi process report NODE_IDQuerying Data with QueryBuilder
Basic query — find failed calculations in a group
from aiida.orm import QueryBuilder, Group, WorkChainNode
q = QueryBuilder()
q.append(Group, tag='g', filters={'label': 'castep-nmr/relax'})
q.append(Node, tag='structure', with_group='g')
q.append(WorkChainNode, tag='w', project=['*'], filters={'attributes.exit_status': {"!=": 0}})Filter by code
from aiida.orm import Code
q.append(Code, project=['*'], with_outgoing='w', filters={'id': 1632})Display results
res = q.all()
for s, w, c in res:
print(s.get_pymatgen().composition, w, c.computer.label)Reference: QueryBuilder tutorial
Fixing Failed Calculations
Simply resubmit with corrected parameters. AiiDA keeps the full provenance — the failed run stays in the graph, and the new run links to it.
🤖 AI Agent for Job Management
This is where AI agents really shine. Instead of memorizing QueryBuilder syntax or verdi flags:
Example prompts for your AI agent:
| What you want | What to tell the agent |
|---|---|
| Check failed jobs | ”Show all failed AiiDA jobs from the past 7 days” |
| Debug a failure | ”Job 1234 failed, check verdi process report 1234 and suggest a fix” |
| Query results | ”Find all CASTEP relaxation calculations that finished successfully and give me the energies” |
| Resubmit | ”Resubmit job 1234 with twice the walltime” |
| Count jobs | ”How many VASP jobs are currently running on Fornax?” |
Workflow with an AI agent:
- Ask the agent to check your job status
- Share any error messages or
verdi process reportoutput - Get a diagnosis and fix suggestion
- Let the agent write the corrected submission script
đź’ˇ Pro tip: AI agents can also generate complex
QueryBuilderqueries from natural language. Just describe what data you need — the agent handles the syntax.