Your Dashboard's Status column shows where each job is. Long names are cut to 10
characters there, so UploadingInputFiles appears as UploadingI.
| Status | Meaning |
|---|---|
| UploadingInputFiles | The job has just been created and its files are being prepared |
| Assembling Experiment | GAMESS, Psi4 or PySCF: the input is being built |
| WebSubmit | NAMD or TINKER: the files are ready and are being sent to a supercomputer |
| SUBMITTING | Being handed to the cluster |
| SUBMITTED | The cluster has accepted it |
| QUEUED | Waiting for a free spot on the cluster |
| ACTIVE | Running |
| COMPLETE | The program has stopped. Check the Result column: it doesn't mean the calculation worked |
| TIMEOUT | The job reached the time limit you chose and was stopped |
| FAILED | The cluster or ChemCompute stopped the job, e.g. it ran out of memory or a node failed |
| CANCELLING → CANCELLED | You (or a resubmit) cancelled it |
| Denied | The job was refused before it ran; see Why a job is refused |
When GAMESS, Psi4 or PySCF hits a problem in the chemistry (the SCF won't converge, the
charge doesn't fit the molecule, the basis set lacks an element), the program stops
by itself. The job's status is then COMPLETE and its Result is Error. A job
that worked says COMPLETE and Success.
So always look at the Result column, or at the heading on the job's results page,
which ends in Success or Error (Job 1234567 (Jetstream2) Error for GAMESS,
Job 1234567 (Jetstream2) COMPLETE Error for Psi4 and PySCF). Then read the Summary
for the reason, and find it on Common job errors.
FAILED with Result ERROR (in capitals) means the job was stopped from outside: it
ran out of memory, a node failed, or the cluster couldn't take it. These can usually be
fixed by resubmitting; see Resubmit offers.
A job waits in QUEUED until the cluster has room. What helps:
(~3 min). (unknown) means there's no recent data.A job that never makes it to a cluster is resubmitted automatically after a few minutes.
Output takes a minute or so to copy back from the cluster. Wait a couple of minutes and
reload. If the results page says
Job is marked complete, but it doesn't seem to be finished, resubmit after a couple of
minutes.
On the Dashboard, press Cancel Job in the details under the job. It's available while the
job is SUBMITTING, SUBMITTED, QUEUED or ACTIVE. CPU-hours are only charged for the time the
job actually ran.
When a job can be fixed without changing its input, its results page offers to resubmit
it. A button under the error in the Summary tab opens the same options. The offers are:
| What happened | The offer |
|---|---|
| The cluster couldn't take the job (none of it ran) | Resubmit on another cluster |
| It ran out of memory | Resubmit with more cores: each core brings about 2 GB of memory |
| It reached its time limit | Resubmit with a longer time limit, or on a cluster with longer limits. It tells you how long nine in ten similar jobs take |
| It ran out of disk space | Resubmit on another cluster |
Press Show Resubmit Options, check the cluster, cores and time limit (they're filled in
for you), then Resubmit Job. You'll get a new job number. You can't resubmit the same
job more than once every 2 minutes.
No resubmit is offered when the input itself needs fixing (a wrong charge, say). Use the
job's settings to start a new job instead; see Viewing results.
TINKER jobs have no resubmit offer. Submit them again from the form.
If you can't submit, the submission page says why, either in a notice above the form or in
the Server Response box after you press Submit.
| Message (shortened) | What to do |
|---|---|
Job submission is currently unavailable for non-US users and unregistered / non-academic users |
Register or sign in with a school or government email that qualifies; see Which email addresses work? |
Verify your email address to submit jobs |
Click the link in the verification email (check spam), or use Resend Verification Email on the Dashboard |
Your account is marked as outside the U.S. |
Turn off any VPN and keep using ChemCompute from the U.S.; it re-checks every night. If it's wrong, open a support ticket |
You do not appear to be connecting from the United States |
Submit once you're back in the U.S. |
You appear to be connecting through a VPN or proxy, and your browser reports a time zone outside the United States |
Disconnect from the VPN and submit from the U.S. If you are in the U.S., set your computer's time zone to your local one |
ChemCompute has not seen you connect from the United States recently |
You'll see this in a notebook. Sign in to ChemCompute in your browser from the U.S., then run the cell again |
Confirm your email address to submit jobs again |
Your yearly email confirmation is overdue. Use the link on the banner |
Over allocation |
The hours you have left, minus those held (Locked) for jobs still queued or running, don't cover this job's cores and time limit. Use fewer cores or a shorter time limit, or Request More CPU-Hours / Resources on the Dashboard; see What a job costs |
GPU access is not enabled for your account |
Set GPUs to 0 |
… can't use GPUs |
Only some packages use GPUs; set GPUs to 0 |
MP2 is not supported on Jetstream2 using 1 core |
Choose Expanse, Bridges-2 or Delta |
No cluster can run this job right now |
All suitable clusters are down or full. Try again later |
Uploading your own … input file needs an account with full submission access |
Use the guided form. On a workshop or student account, ask for the researcher role; see Asking for more |
Error: The site has been updated … refresh your cache (Ctrl-F5) |
Hard-refresh the page (Ctrl+F5, or Cmd+Shift+R on a Mac) |
Your session could not be verified |
Reload the page and submit again |
Your account does not have permission to submit jobs |
Contact us |
If you choose more cores or a longer time limit than your role allows, the job isn't
refused: it is quietly reduced to the largest allowed setting. Check the Resources on
the job's page to see what it actually got.