source-instagram v4.2.18: intermittent user_insights failure with Meta 400 "(#100) since param is not valid" and follow-up "Concurrent read failure"

[lubimow]

Connector Name

source-instagram

Connector Version

4.2.18

What step the error happened?

During the sync

Relevant information

After upgrading Instagram source to v4.2.18, we started observing intermittent sync failures on the user_insights stream.
Deployment type: self-hosted Airbyte Community (not Airbyte Cloud), running on our own server (platform version 2.0.1).

The failing runs return Meta API error:
(#100) since param is not valid. Metrics data is available for the last 2 years

After that source-side error, the sync often ends with:
Concurrent read failure
and destination-side transient errors about missing terminal stream status messages.

What is unexpected

• The same connection can sometimes succeed and then fail again on the next scheduled/manual run.
• The error appears stream-specific (user_insights), while other streams may still emit records.
• This looks like unstable window/state behavior around since/until generation for user_insights.

Connector/stream context

• Source: Instagram (4.2.18)
• Destination: BigQuery
• Stream: user_insights (incremental)
• Sync mode: Incremental | Append
• Cursor field: date
• Server timezone observed: Europe/Kyiv (EET, +0200), system clock synchronized

Error details (redacted)

Example failing request pattern from logs:
GET //insights?period=day&metric=follower_count,reach&since=2026-02-26T08:00:00+00:00&until=...

Returned error:
HTTP 400: (#100) since param is not valid. Metrics data is available for the last 2 years

Then:
AirbyteTracedException: Concurrent read failure

And destination side:
TransientErrorException: Input was fully read, but some streams did not receive a terminal stream status message

Steps to reproduce (observed behavior)

1. Configure Instagram -> BigQuery connection with user_insights enabled in incremental mode.
2. Upgrade source connector to v4.2.18.
3. Run sync multiple times (manual + scheduled).
4. Observe intermittent failures with Meta 400 on user_insights and subsequent concurrent read failure.

What we already tried

1. Refreshed source schema.
2. Verified stream configuration (user_insights incremental append, cursor date).
3. Updated source start_date to a recent value within allowed range (example: 2024-05-26T00:00:00Z).
4. Re-ran syncs multiple times.
5. Verified server time/timezone (Europe/Kyiv, clock synchronized).

Result after mitigations

• Some runs complete successfully.
• Failures still reappear intermittently with the same user_insights 400 error.

Request

Please confirm:

1. How since/until are computed internally for user_insights (timezone/day-boundary logic).
2. Whether this is a known regression in/after v4.2.18.
3. Recommended safe workaround that does not require clearing destination data.
4. Whether a specific connector version contains a fix.

Log snippets to include

• Source error with (run tests in default docker mountpoints #100) since param is not valid...
• Concurrent read failure
• Destination transient error about missing terminal stream status messages

Relevant log output

- instagram___ile_bigquery_logs_1475_txt.txt:96 
  ... GET https://graph.facebook.com/v23.0/<ig-business-account-id>/insights
  ?period=day&metric=follower_count,reach
  &since=2026-02-26T08:00:00+00:00&until=2026-02-27T00:58:47+00:00
  -> HTTP 400: (#100) since param is not valid. Metrics data is available for the last 2 years

  - instagram___ile_bigquery_logs_1475_txt.txt:371 
  TransientErrorException: Input was fully read, but some streams did not receive a terminal stream status message
  Streams without a status message: [instagram_user_insights, instagram_user_lifetime_insights, instagram_media, ...]

  - instagram___ile_bigquery_logs_1480_txt.txt:4101 
  ... same Meta 400 on user_insights with since=2026-02-26T08:00:00+00:00 ...

Contribute

• Yes, I want to contribute

---

Internal Tracking: https://github.com/airbytehq/oncall/issues/11268

[octavia-bot]

AI Triage Starting...

Thank you for opening this issue! I'm an AI assistant that will do an initial review and may:

• Ask clarifying questions if needed
• Suggest relevant documentation or workarounds
• Flag this for internal visibility

Note: Community issues are reviewed on a best-effort basis and do not have a guaranteed response time. For additional help:

• Join the Airbyte Community Slack to connect with other users and community members
• If you're an Airbyte Cloud customer, you can open a support ticket or contact your account representative, referencing this issue URL

View playbook

• 🔗 Session URL: https://app.devin.ai/sessions/c6cf61218b874400b046bd6257d5e09a

✅ Devin AI session created successfully!

[devin-ai-integration]

Hi romo (@lubimow), thank you for reporting this issue with such thorough detail!

This looks like a known problem related to how source-instagram computes since/until parameters using UTC, while the Meta API interprets day boundaries based on the account's local timezone. Since you're in Europe/Kyiv (UTC+2), the UTC-derived timestamps can fall outside the valid 2-year window from Meta's perspective, causing the (#100) since param is not valid error intermittently.

The "Concurrent read failure" and destination-side "missing terminal stream status messages" errors are downstream consequences — when user_insights fails mid-sync, the platform cannot finalize the stream status properly.

What we know

• This is a confirmed pattern affecting multiple connectors that use UTC-based date computation against APIs that interpret dates in the account's local timezone.
• source-instagram is listed as a moderate risk connector for this issue because the IG User API does not directly expose a timezone field, making a programmatic fix more complex.
• A fix has already been applied to source-facebook-marketing (airbytehq/airbyte#73281) for the same root cause pattern.

Workaround

As a temporary mitigation, you can try adjusting your start_date to ensure the computed since value stays well within the 2-year window. However, since the issue is intermittent and depends on timing relative to your timezone offset, this may not fully resolve it.

This issue is being tracked internally: https://github.com/airbytehq/oncall/issues/11268

---

Need more help? Join Airbyte Community Slack for peer support, or if you're a Cloud customer, open a support ticket referencing this URL.

This message is generated from Devin.ai (internal session link).