Troubleshooting
Find what an error or refusal in eode means and what to do about it, from RIA stalling to the engine being down.
Where: RIA diamond→the red text in the console
Most errors in eode say what to do in the message itself. This page collects them in one place, with what each one means. Errors that belong to one screen are listed on that screen's page; the table at the end points to them.
When RIA stops or fails
Console errors appear in one box at the foot of the conversation. A technical code and some details may follow the message; they're for River Logic support. There's no Retry button: in every case, send your message again. Sending the same message twice never runs it twice.
| The console says | Means | Do |
|---|---|---|
| RIA stopped responding — the connection dropped or the turn stalled. Send your message again to retry. | The answer went quiet, and the one automatic reconnect found nothing. | Send again. |
| Your engine didn't start | Your own optimization engine failed to start. The message ends Send your message again to retry. | Send again. |
| The console lost track of your engine's startup | The spinning-up line ran for 150 seconds with no word back. The message ends refresh or send your message again to retry. | Reload, or send again. |
| your engine is starting — retry shortly | Your engine is on its way up. | Wait a few seconds, send again. |
your engine did not become ready within <n>s | Your engine took too long to start. | Send again. |
| An engine-unavailable message | Your engine can't be reached at all. | Wait a minute, send again. If it persists, see the engine is down. |
This turn stopped responding and was canceled after <minutes> minutes. The engine has been freed — please try again. | The answer ran past 40 minutes and was stopped. | Send again, asking for less. |
| The server restarted while RIA was working on this, so the turn did not finish. Your conversation is saved — send your message again in a moment to retry. | eode restarted mid-answer. | Wait a moment, send again. |
| This turn failed before RIA could start working on it. Nothing was changed — please send your message again. | It failed before anything happened. | Send again. |
| I couldn't finish that — please try again. | The answer failed partway through. What went wrong is kept for River Logic support, not shown. | Send again. |
| The server reported an error without any detail. Your session is still alive — send the message again to retry. | An unexplained failure. | Send again. |
| A message that another answer is still in progress | Another tab of yours is mid-answer in the same conversation. | Wait a few seconds, send again. |
| A message that the service is at capacity | Too many answers are running at once. | Wait a few seconds, send again. |
| A message that the conversation isn't yours | The conversation belongs to someone else. | Open one of your own from the conversation menu. |
| your access was revoked / authentication required | You've been signed out, or removed from the organization. | Sign in again. |
When RIA pauses an answer
A long answer can stop partway, with what RIA has so far above. These are RIA's own replies, not errors:
| RIA says | Do |
|---|---|
| That took longer than expected, so I paused — what I have so far is above. Tell me to continue and I'll pick right back up. | Reply "continue". |
| The engine had to reload your model partway through, so this took longer than it should and I paused. What I have so far is above. Tell me to continue and I will pick up where I left off. | Reply "continue". |
| I had to stop before finishing this answer. … Ask me again, and if it stops again, ask for less in one message. | Ask again, or split the question. |
While your model reloads, the spinner line reads The engine had to reload your model before it could answer. That takes a moment.
When RIA says it can't
Some refusals come back as RIA's reply:
| RIA says | Do |
|---|---|
| No model is active — open one first. | Open a model, then ask again. |
| This model's structure isn't editable yet. | Imported models and older builds can't be changed by RIA. See Change a model. |
| The user doesn't have permission to change this workspace. | Ask the workspace's owner for edit access. |
| The user doesn't have permission to read this workspace's model. | Ask the model's owner for access. |
When the engine is down
The header shows a red dot and ENGINE · DOWN at the right (Window tour). If the engine keeps failing its checks, a banner under the header also reads Optimization engine unavailable. Solves can't run on this environment right now. If you need one, contact support. Both go away by themselves when the engine is healthy again; eode checks every 10 seconds.
When you're sending too fast
Too many requests - limit 30 per 60s. Retry in <n>s. means you've hit the fair-use limit: 30 per
minute per person, counted separately for console messages, solves and AI model-building. Wait the
seconds it names, then try again.
When a what-if goes to RIA instead
This reads as a question for RIA rather than a what-if. Ask it in chat, or add an override to make it a scenario. A what-if you wrote reads as a question and changes nothing, so it's sent to RIA in the console instead. To make it a scenario, add an override (What-ifs).
Things that are normal
- No pop-ups. eode never shows toasts or browser notifications. Each confirmation is a line next to
what you did, such as Invitation sent to
<email>. - Confirm before anything permanent. A confirm dialog has Cancel and OK, or a red button like Delete or Remove for something that can't be undone.
- The page breathes. While RIA works, the dark theme's background pulses slowly. Motion in Page settings set to reduce turns it off.
- The diamond pulses while RIA works, or while a model registers or uploads.
Errors on other pages
| Where | See |
|---|---|
| Sign-in pages: Sign-in didn't complete, Your account is not in an organization, Your membership could not be completed yet, and the page saying the service can't be reached | First run |
| Dropping a file on the console | The RIA console |
| A proposal or change order that won't apply | RIA's proposals |
| Inviting, removing or renaming in your organization | Your organization |
| A help request that didn't send | Help and support |
| Promote or Verify refused on a candidate | Review Queue |
Limits
- Fair use: 30 per minute per person, each for console messages, solves and AI model-building.
- A console answer is stopped after 40 minutes. RIA pauses a chat answer after 4 minutes and a model-building one after 15.
- After 150 seconds of waiting for your engine to spin up, the console stops waiting and says so.
- The engine's health is checked every 10 seconds.
Related
- Help and support: report what isn't on this page
- Limits: every limit in eode