All I am getting is this: "I hit a problem reaching the AI service mid-task: AI Troubleshooting is temporarily unavailable. Your box itself is fine — everything in your dashboard (apps, logs, updates) keeps working while we sort this out. Try me again in a minute; if it keeps repeating, contact support. (code U402). Try again in a moment." Restarted everything. Its been like this for about an hour.
13 replies
javielitou wrote:
I was getting the same message when configuring. It just forced me to figure it out. LOL
Chris wrote:
You caught a real one — that U402 was our side, not yours. Our AI provider hit an account problem shortly after midnight and Tom AI stopped answering for about 90 minutes. We have switched it back to its main provider and verified it is answering as of a couple of minutes ago — try again and it should work normally. Sorry for the late-night runaround, and thanks for posting; this thread is what got it spotted and fixed.
iBangax132 wrote:
still getting this error: I hit a problem reaching the AI service mid-task: request failed. Try again in a moment. any other news why this is happening?
Chris wrote:
Thanks for reporting it here — that 'request failed' is a different, more generic error than the U402 from overnight, and both are on the AI service side rather than anything on your box. The team is already watching the service today (the U402 was a provider account problem we moved off), so I'll pass this fresh one straight through to them. Everything else on your box keeps working normally in the meantime. Sorry for the back-and-forth — give it another try shortly, and if it still won't connect I'll keep it flagged.
iBangax132 wrote:
was happening this morning and still consistent errors, checked just now - so been about 5-6 hours of it (was working yesterday no problem)
Chris wrote:
Five or six hours of this is far too long, and I'm sorry it's still down — you shouldn't be the one chasing it. This is on our side, not your box: your apps, logs and updates all keep working while we sort the AI service. I've re-flagged it to Tom with your timing just now so he can chase the provider directly. I'll post back here the moment it's back. Nothing for you to do in the meantime.
Chris wrote:
Good news on this one: Tom shipped an update today that makes Tom AI say what's actually wrong instead of a bare 'request failed' — that generic message was hiding the real cause from us too. Update SparkBox with: sudo sparkbox upgrade — then try Tom AI again. If it still errors, the message will now show the actual reason (or a code), and pasting that here will let us pin it down immediately.
iBangax132 wrote:
currently on Current version v1.6.558, updated it last night and AI still isnt working although i hate updating it as it removes my mount and my jellyfin then has to scan 20TB of data again. wonder if Tom can fix the mounting issue at some point
Chris wrote:
That version number is the giveaway — 1.6.558 is yesterday's release. The fix that makes Tom AI actually say what's wrong (instead of a bare 'request failed') landed today in 1.6.563, and 1.6.564 is current. To pick it up run sudo sparkbox upgrade — update only refreshes app images, upgrade is the one that moves you to a new release. On 1.6.563+ it'll either connect or finally name the real reason, and we can chase it from there. On the mount: a mount hand-edited into the compose file does get dropped on upgrade. The durable home for it is a small per-app override file under /opt/sparkbox/state/overrides/ (media bundle = media.yml) — that survives upgrades, and 1.6.564 now warns when it spots a hand-edited compose file and points you there.
iBangax132 wrote:
upgrading the box fixed the ai issue that is now working, so i now shouldnt be worried about upgrading and losing my mount?
Chris wrote:
Really glad the upgrade got Tom AI working again — and no, you shouldn't have to worry about losing that mount every time. Here's what's going on: SparkBox rebuilds its app config on every update, so a mount added by hand (or through Portainer) gets wiped — that's what forces Jellyfin to rescan the 20TB. The version that sticks is SparkBox's own Media Root setting, in the configuration under the Media section. Point it at the folder your library lives in, then run: sudo sparkbox up. After that Jellyfin keeps its library across updates. If you've already set Media Root there and it still vanishes, tell me — that would be a real bug, not setup. And making that media-folder setup easier is already on Tom's list.
iBangax132 wrote:
TomAi has gone down again: I hit a problem reaching the AI service mid-task: ETIMEDOUT. Try again in a moment.
Chris wrote:
Thanks for the fast flag — that 'ETIMEDOUT' is a new, different error from the two earlier today, and it's the AI service timing out on our side, not your box. I've flagged this fresh one for Tom with the exact wording, so there's nothing you need to do; your apps and library keep running normally while he chases it.