Hi everyone, I’m seeing an inconsistency between t...
# airbyte-agents-ask-ai
s
Hi everyone, I’m seeing an inconsistency between the PostgreSQL and MSSQL source connectors when syncing empty tables with Full Refresh | Overwrite. Environment: • Airbyte:
0.58.0
(OSS/Docker) • MSSQL:
source-mssql:4.0.15
• PostgreSQL:
source-postgres:3.8.1
• Destination:
destination-clickhouse:1.0.0
With PostgreSQL, an empty table is still included in
streamStats
with
recordsEmitted: 0
,
bytesEmitted: 0
, etc. With MSSQL, the sync succeeds, but an empty table is completely absent from
streamStats
. Only non-empty tables are reported. For example, with 3 tables (50k, 50k, and 0 rows): • PostgreSQL → all 3 streams appear in
streamStats
• MSSQL → only the 2 non-empty streams appear Is there a recommended way to identify that an empty MSSQL table was successfully processed?
k
This is a known inconsistency that has been reported and escalated to the Airbyte team. Here's what's known: MSSQL Empty Stream Reporting Gap The behavior you're describing is confirmed:
source-mssql:4.0.15
does not include empty streams in
streamStats
, while
source-postgres:3.8.1
does. The Airbyte team has escalated this for investigation (internal tracking:
airbytehq/oncall#13252
) but has not yet confirmed whether this is intentional or a bug, nor has a fix been committed. [MSSQL streamStats discussion] Interim Workaround Until the behavior is resolved, the recommended approach from the Airbyte maintainers is: > Treat the configured catalog as the source of truth instead of
streamStats
.
Specifically: 1. Retrieve the full set of enabled streams for the connection from the connection's configured catalog (available via the Airbyte API). 2. After a sync completes, compare that set against the streams present in
streamStats
. 3. For any successful job (
status: succeeded
), treat any enabled stream with no
streamStats
entry
as "processed with zero records." ⚠️ Important caveat: This inference is only safe when the overall job status is
succeeded
. A failed or cancelled job can also omit streams from
streamStats
for unrelated reasons, so you must not conflate the two cases. [MSSQL streamStats discussion] Additional Context the Team Has Requested If you want to help the investigation, the maintainers have asked for: [MSSQL streamStats discussion] • Whether the same omission occurs on an Incremental sync of an empty table (not just Full Refresh | Overwrite). • Relevant source connector log excerpts for the empty stream from an MSSQL sync (stream status or state lines), with sensitive info masked. • Whether the empty table is still created/emptied correctly in ClickHouse despite being absent from
streamStats
. There is currently no public roadmap item or existing GitHub issue specifically tracking
streamStats
consistency for empty streams across database sources. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
s
@kapa.ai Could you share the internal tracking link for this issue?
k
Based on the knowledge sources, the internal tracking link for this issue is
airbytehq/oncall#13252
, referenced in the maintainer's comment on the discussion. [MSSQL streamStats discussion] Note that this is an internal Airbyte repository (
airbytehq/oncall
) which is not publicly accessible, so you won't be able to view the issue directly. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
s
@kapa.ai Any update on this? I have also added reply in the github dicussion, so please notify to corresponding person.
k
I'm sorry, but I don't have any updates beyond what's already been shared — the knowledge sources only show that the issue has been escalated internally (
airbytehq/oncall#13252
) with no official answer yet. [MSSQL streamStats discussion] Regarding notifying the corresponding person — I'm an AI assistant and am not able to send notifications, tag people, or interact with GitHub on your behalf. To get attention on your reply, you could: • Tag the maintainer who responded in the GitHub discussion directly in your comment. • Join the Airbyte Community Slack and raise it there for peer support or maintainer visibility. • If you're an Airbyte Cloud customer, open a support ticket referencing the discussion URL. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
s
@kapa.ai Who is maintainer for github?
k
The knowledge sources do not contain specific information about who the maintainer is for the GitHub discussion you're referring to. The maintainer's response in the discussion is posted anonymously without a named GitHub handle visible in the knowledge sources. [MSSQL streamStats discussion] I'd recommend: • Visiting the GitHub discussion directly to see the maintainer's username and tag them in your reply. • Reaching out via the Airbyte Community Slack for quicker visibility. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
s
@kapa.ai any further update on this? Seems like it is stuck and no one is working on this.
k
I understand your frustration, but unfortunately the knowledge sources do not contain any further updates beyond what was previously shared. The discussion still shows no official answer from the Airbyte team. [MSSQL streamStats discussion] To help get this moving, here are a few suggestions: • Comment directly on the GitHub discussion asking for a status update — visibility and upvotes from the community can help prioritize issues. • Raise it in the Airbyte Community Slack where maintainers are often more responsive. • Open a support ticket at support.airbyte.com if you're an Airbyte Cloud customer, referencing the discussion URL. In the meantime, the interim workaround of using the configured catalog as the source of truth (treating any enabled stream absent from
streamStats
in a
succeeded
job as "processed with zero records") remains the best available approach. [MSSQL streamStats discussion] 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.
s
@kapa.ai Could you provide the maintainer name?
k
I'm sorry, but the provided knowledge sources do not explicitly mention the specific name or GitHub username of the maintainer who responded to that discussion. The response in the document is only attributed generically as "Comment 1 by maintainer". [MSSQL streamStats discussion] You can view the specific username by checking the GitHub discussion directly. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups or DM the app for 1-to-1 help.