Fix HubSpot delivery when it is paused

HubSpot delivery pauses when the portal returns an error that another attempt cannot fix. The HubSpot page under Integrations in the sidebar (/integrations/hubspot) then reads “Delivery to HubSpot is paused” at the top, and Needs attention under it shows the reason with Resume.

Read the pause reason

Authentication errors pause on the first occurrence, because a retry sends the same token. A request HubSpot refuses outright pauses on the first occurrence too, because sending the identical body again produces the identical rejection.

One contact refused for its own values does not pause anything. A 400 VALIDATION_ERROR (an invalid email, a value a portal rule rejects), a 409 CONFLICT (the contact already exists), or a lead with no properties to write fails that delivery only, and Embers sends the next lead. Only five of those in a row pause the install. One synced contact resets the count. Resuming HubSpot, switching delivery back on, or reconnecting starts the count from zero, so HubSpot pauses again only after five more refused contacts in a row.

If a contact Embers synced is deleted or merged in HubSpot, Embers creates it again (or matches an existing contact by email) on the next update and keeps syncing. HubSpot is not paused. If HubSpot still cannot find the contact, only that lead’s delivery fails, and you can choose Retry on its row on the Activity card.

Rate limits (429), timeouts, and 5xx errors are retried instead: the first attempt plus seven more, from 30 seconds out to about 6.3 days, the same schedule as webhook retries.

The token lacks a scope

Embers creates and updates contacts, so the private app needs both crm.objects.contacts.read and crm.objects.contacts.write. Connecting only checks that the token can read the portal, so a read-only token connects and then pauses on the first real lead. HubSpot reports a missing scope as a 403, which is why it shows as INVALID_AUTHENTICATION rather than a scope error.

A property was refused

Embers writes only firstname, lastname, jobtitle, company, and website, plus email and phone when a paid plan has them stored. Empty values are left out rather than sent as blanks. Five refused contacts in a row therefore usually point at a validation rule or a changed property type in your portal. The full field list is in what Embers writes to HubSpot.

Fix and resume

Recover the leads you missed

Leads that qualified while HubSpot was paused were kept as Blocked, not sent. After you resume, choose Retry on a blocked row on the Activity card, or select the leads on the leads list and send them to HubSpot. A pushed lead that already has a linked HubSpot contact is skipped as already synced to HubSpot contact unless you tick ignore rules, in which case the existing contact is updated rather than duplicated.

Troubleshooting

• The pause returns right after reconnecting. The new token still lacks a scope. Check both contact scopes, not only read.
• Save token says HubSpot rejected the token. The token is wrong or revoked: HubSpot rejected that private app token. Check crm.objects.contacts read and write scopes. Copy it again from the private app.
• Duplicate contacts appeared. A lead without a stored email cannot be matched to an existing contact by email. See how duplicates are avoided, then merge them in HubSpot.
• Resume is missing. Delivery is not paused, or you are not an account administrator.

Send the pause reason and the time to viraj@useembers.com if it does not clear. Slack follows the same pause rules: see Slack delivery is paused.

Related