@kapa.ai why one of my connection, connection state : [ { "streamDescriptor": { "name": "keywords_report" }, "streamState": { "state": { "date": "2026-02-17" }, "parent_state": {}, "lookback_window": 18, "use_global_cursor": true } } ] ; is always "date": "2026-02-17" even I refresh, others are all the date when it refreshed?

Looking at your connection state, your

keywords_report

stream is using

"use_global_cursor": true

, which means it's operating under the Global Substream Cursor mode. This is the key reason your date stays fixed at `"2026-02-17"`: > The global cursor is updated only at the end of the sync. If the sync fails, only the parent state is updated — the global cursor itself is not updated. [Incremental Syncs - Nested Streams] So if your syncs are failing before completion, the global cursor (

date

) will remain stuck at its last successfully committed value (

2026-02-17

), while other streams (which likely use per-partition cursors) update their state independently as records are processed. Additionally, the global cursor enforces a lookback window (you can see

"lookback_window": 18

in your state) to avoid missing records added during a sync. This is by design to ensure data completeness. What you can do: 1. Check if your syncs are completing successfully — if the sync is failing partway through, the global cursor won't be updated. 2. Review the connection state manually via the Airbyte UI: ◦ Go to Connections → select your connection → Settings tab → Advanced dropdown → Connection State. ◦ You can edit the state to manually update the date if needed, but do so with caution. [Manage Connection State] > ⚠️ Updates to connection state should be handled with extreme care. Changes may break your syncs, requiring a full historical sync to fix. Make changes only as directed by the Airbyte team. [Manage Connection State] 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai What I gave you is already Review the connection state manually via the Airbyte UI: Go to Connections → select your connection → Settings tab → Advanced dropdown → Connection State. but the syncs are all successful

Thank you for clarifying. So your syncs are completing successfully, but the

date

in the

keywords_report

stream state remains stuck at

"2026-02-17"

. The knowledge sources I have available don't contain enough specific information to definitively explain why a successfully completing sync would leave the global cursor date unchanged for

keywords_report

while other streams update normally. However, there is one relevant piece of information: the

keywords_report

stream (or a similarly named stream

keywords_report_daily

) in Apple Search Ads had a known migration that specifically addressed cursor behavior: > This release introduces changes to incremental sync of

keywords_report_daily

streams, changing the stream state from per-partition state to a global state cursor. [Apple Ads Migration Guide] The recommended fix in that migration was to clear the stream data and re-sync: 1. Go to Connections → select the affected connection. 2. In the Status tab, find the affected stream in the Enabled streams list. 3. Click the three dots on the right side of the stream and select Clear Data. 4. After the clear succeeds, trigger a sync by clicking Sync Now. [Apple Ads Migration Guide] Beyond this, the knowledge sources don't provide enough detail to fully explain your specific situation. I'd recommend reaching out to the Airbyte team directly or opening a GitHub issue, as this may be a platform-level bug similar to the one described in this open issue where stream state is saved but not properly updated on subsequent syncs. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai Recent data airbyte_generation_id are 2 instead of 1, is it related ?

Based on the conversation history, you previously ran a Clear on the

keywords_report

stream (or it was cleared as part of a refresh), and then a sync ran afterward. This is directly explained in the Airbyte documentation: > The generation ID will be incremented whenever you run a clear or refresh. Airbyte will never decrease the generation ID. If you run a Refresh immediately after running a Clear, that will also increment the generation ID. This means the generation ID will increment twice, even though one of those "generations" never emitted any records. [Refreshing your data] So yes, this is likely related. If a Clear was run (either manually or as part of a migration/refresh), the generation ID incremented once (to 1), and then the subsequent sync incremented it again (to 2). This is expected behavior — it does not indicate a problem with your data. To confirm: •

_airbyte_generation_id = 0

→ data from before any clear/refresh •

_airbyte_generation_id = 1

→ the clear incremented it (no records emitted in this generation) •

_airbyte_generation_id = 2

→ the actual re-sync after the clear This is normal and by design. The

_airbyte_generation_id = 2

records are your current, valid data. [Refreshing your data] 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai No

_airbyte_generation_id = 2

are data after refresh. My question now is why I could only fetch very few data

The knowledge sources I have available don't contain enough specific information to definitively explain why you're fetching very few records after your refresh, especially in the context of your specific

keywords_report

stream state being stuck at

"2026-02-17"

. However, there is one potentially related known bug documented in the knowledge sources: > After a truncate refresh, raw tables are correctly populated with all records, but final tables are overwritten with empty

_airbyte_tmp

tables because the T+D

generation_id

filter matches 0 rows — and the sync still reports success. [Truncate refresh bug] This could be relevant if you're seeing very few or no records in your final tables despite the sync appearing successful. The bug describes a scenario where: • Raw tables are correctly populated • Final tables end up nearly empty due to a

generation_id

mismatch in the Type and Deduplicate step • The sync reports success regardless What you can check: 1. Compare the record count in your raw tables vs. your final tables — if raw tables have the expected data but final tables don't, this bug may be affecting you. 2. Check which version of Airbyte you're running. The issue was reported on version 1.6.7, and it's unclear if it has been fully resolved in later versions. I'd strongly recommend opening a GitHub issue or contacting the Airbyte team directly with your specific version and logs, as the knowledge sources don't provide enough detail to fully diagnose your situation. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai It's not related to especially in the context of your specific

keywords_report

stream state being stuck at

"2026-02-17"

.

Understood, thank you for the correction. So to summarize the situation: • Your

keywords_report

stream state is stuck at

"2026-02-17"

(separate issue) • Your

_airbyte_generation_id = 2

is expected and normal after a clear + re-sync • Your current concern is: why are you fetching very few records after the refresh? Unfortunately, the knowledge sources I have available don't contain enough specific information to explain why only a small amount of data is being returned after your refresh for the

keywords_report

stream. There is one potentially relevant known bug worth checking: > After a truncate refresh, raw tables are correctly populated with all records, but final tables are overwritten with empty temp tables because the

generation_id

filter in the Type and Deduplicate step matches 0 rows — and the sync still reports success. [Truncate refresh bug] What I'd suggest checking: 1. Compare record counts in your raw tables vs. your final tables — if raw tables have the expected volume but final tables don't, this bug may be affecting you. 2. Check your Airbyte platform version, as this bug was reported on 1.6.7. Beyond this, I'd recommend reaching out to the Airbyte team directly or opening a GitHub issue with your platform version, connector version, and sync logs for a more definitive diagnosis. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai I am only talking about raw tables, and it become very less. Airbyte platform version is 2.0

The knowledge sources I have available don't contain enough specific information to explain why your raw tables would have very few records after a refresh on Airbyte platform version 2.0. The known bug documented in the sources [Truncate refresh bug] specifically describes the opposite scenario — raw tables are correctly populated but final tables end up empty. So that bug does not apply to your situation. Given that: • Your raw tables have very few records after a refresh • Your syncs report as successful • Your

keywords_report

stream state is stuck at

"2026-02-17"

The stuck cursor date could actually be relevant here: if the stream is using

"use_global_cursor": true

and the cursor date is stuck at

"2026-02-17"

, the source connector may be filtering and only sending records newer than that date — which, depending on your source, could result in very few records being written to the raw tables. However, I want to be transparent: the knowledge sources don't provide enough detail to confirm this definitively for your specific connector and setup. I'd recommend: 1. Checking whether the record count in raw tables aligns with what you'd expect from records newer than

"2026-02-17"

2. Opening a GitHub issue or contacting the Airbyte support team with your connector name, platform version (2.0), and sync logs for a proper diagnosis 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai No,

keywords_report

stream state is not stuck

Thank you for the correction. To recap what we know: • Your

keywords_report

stream state is not stuck — it updates correctly after refreshes • Your

_airbyte_generation_id = 2

is expected and normal • Your syncs report as successful • Your raw tables have very few records after the refresh on Airbyte platform version 2.0 Unfortunately, the knowledge sources I have available do not contain enough information to explain why your raw tables would have very few records after a successful refresh in this scenario. I'd recommend: 1. Checking your source — verify that the source system actually has the expected volume of data available for the sync window 2. Reviewing the sync logs in detail for any warnings about skipped or filtered records 3. Contacting Airbyte support or opening a GitHub issue with your connector name, platform version (2.0), connector version, and sync logs for a proper diagnosis, as this goes beyond what the available knowledge sources can address 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai What if disabled the current one and create a new connection of the same configuration? Is it feasible ?

Yes, this is feasible. Here's what you need to know: Disabling/deleting the current connection: • You can delete the existing connection via Connections → select the connection → Settings → Delete this connection • Note: deleting is irreversible, but your data in the destination remains intact [Manage Connections] Creating a new connection with the same configuration: • You can create a new connection using the same source and destination • However, be aware that reestablishing a deleted connection requires a full re-sync [Manage Connections] To avoid a full re-sync (for incremental streams): You can copy the state from the old connection to the new one: 1. Before disabling/deleting, go to the old connection → Settings → Advanced → Connection State and copy the state 2. On the new connection, paste the copied state and click Update State 3. This ensures historical data is not replicated again [Upgrading Connections] > ⚠️ Updates to connection state should be handled with extreme care. Updates may break your syncs, requiring a full historical sync to fix. [Manage Connection State] This approach is a reasonable troubleshooting step, but whether it resolves your raw table record count issue is something the knowledge sources cannot confirm — it depends on the root cause of the problem. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.

@kapa.ai is it because I drop fields (insights or keywordId or other) that cause incremental | append not working ? something like lost the unique key

Sorry @Yenying Chen, I'm currently experiencing some technical difficulties. Please try again in a few minutes. If this issue persists, please reach out to the administrators of this Slack workspace.