FAQ and troubleshooting

A post didn't go out. Why?

Open the item in the app and read its state, then check in this order (the phone app):

  1. Needs approval or Re-approve: nothing goes out until you approve it with your passkey, and any change to its texts or media needs a new approval.
  2. On hold: you held it; it waits until you press Release hold.
  3. Publishing is paused. (the red banner): approved items wait until you Resume with passkey.
  4. Missed by more than 6 hours (CATCH_UP_S, for instance the instance was down): it isn't published late. The notification says <key>: held, "not published within 6 h of its time"; post it by hand, or move the date and approve again. After a pause, overdue items go out only within that window.
  5. Failed. The notification's title is <key>: failed and its text says why: <platform> reported: <error>, or " didn't finish processing within 2 h". Bongsu tries a step three times (and waits for Instagram's processing up to 24 times) and then stops. Approving it again starts it over from a new upload (an approval made after the failure; nothing an agent does reopens it). The exception: once Bongsu has asked the platform to publish, the post may be live even though the call failed, so approving again doesn't reopen it, to never post twice. Check the account, then record it with record_posted (API reference) or post it by hand.

"A connection needs attention"

A token expired or was revoked, and nothing is posted for that account until it's reconnected. Instagram tokens last 60 days and Bongsu renews them every week; a renewal fails when the Instagram password changed or access was removed in Instagram. YouTube's refresh token is checked weekly; if the Google app is still in Testing, Google expires it after 7 days, so publish the app. Go to the dashboard's Connections and Reconnect: Instagram, YouTube. With a mailer set, the workspace's admins also get an email, "Bongsu: a connection needs attention".

Instagram refuses a post

  • Limit: 100 published posts per account per 24 hours.
  • The file can't be fetched: Instagram downloads the media from Bongsu's signed link, so PUBLIC_URL must be reachable from the internet over HTTPS (a local instance needs a tunnel: Instagram).
  • Never finishes: Bongsu waits for Instagram to process the video, checking every 5 minutes up to 24 times (2 h), then fails the item: "Instagram didn't finish processing within 2 h" (the platform's name is the one Bongsu shows). Check the format (MP4/MOV, H.264) and approve again: that starts it over with a new upload.

YouTube says the quota is gone

The project gets 10,000 units a day; a comment or reply costs 50. Bongsu posts at most two Shorts comments per run, so a few channels stay far below it; it resets daily (YouTube).

"No media in the store yet: it can't be approved until the file is there."

The item's files aren't where its path says, in the default media store (the dashboard's Media page shows which one that is). On Drive, check the folder names against Google Drive (a renamed folder loses its items' media); in a bucket, the keys (S3-compatible buckets); on the server's disk, the files must be under MEDIA_DIR at the same paths (the bongsu-media volume on Docker). A replaced file changes its checksum and needs approving again.

The times are off

The database keeps UTC; the zone is only for showing and entering. The workspace's zone is the default, and Settings → Time zone in the app sets yours; changing a zone moves nothing already planned. A new instance starts on UTC (Configuration). An agent reads due (on its workspace's clock) and due_at (UTC) (API reference).

The app says a newer version is deployed

The installed app is an older copy. Settings → Version → Update now reloads it (Updating).

I got a new phone

Passkeys belong to a device. From the dashboard's Devices & passkeys, Add a device, scan the code on the new phone and create its passkey. If you lost every passkey, a recovery code signs you in once; Make new codes afterwards (The dashboard).

"SETUP_TOKEN isn't set on the Worker (site/docs/deploy-cloudflare.md)"

The first visit to the dashboard asks for SETUP_TOKEN, and the instance doesn't have one (on Docker the message still says Worker). Set it as a secret (bongsu deploy-secrets generates it and prints it once, or wrangler secret put SETUP_TOKEN; in .env on Docker), deploy or restart, and try again (Deploy on Cloudflare).

Edit this page