Hey folks 👋 Just submitted my first connector PR for the ComfyUI Cloud API (the open-source AI image/video generation platform with 4M users, adopted by Netflix, OpenAI, Amazon Studios). Key Features: - 6 streams:

jobs

,

job_details

,

assets

,

models

,

nodes

,

system_stats

- Incremental sync: Supported on

jobs

and

assets

- 71 unit tests - Validated against the real Cloud API This is the first generative AI workflow engine connector in the Airbyte ecosystem. I think there’s real demand, as creative studios are scaling up AI generation pipelines and need observability over costs, GPU usage, and model performance. - PR: https://github.com/airbytehq/airbyte/pull/78142 - Discussion: https://github.com/airbytehq/airbyte/discussions/78141 Happy to iterate on any feedback! First contribution for me, so I'm excited to be here. octavia wave

Hi @Henry Mattingly! I submitted a ticket request for the Square connector 3 weeks ago and it's just waiting on a final human review! Would you be able to share the steps left to having it go live? I would like to use Airbyte to solve this problem and avoid rebuilding it custom on my own. https://github.com/airbytehq/airbyte/pull/77704#issuecomment-4415221037

Hi Airbyte community — I’m running Airbyte v2 OSS in a fully air-gapped environment and hitting a 500 Internal Server Error on the connector documentation endpoint. In the browser console, the UI fails while loading destinations and shows connector_documentation/get errors, with repeated 500 responses �. I’m using a VPN-configured machine with Chrome on the same machine, and I’m trying to connect from there. Once the VPN is connected, there is no internet access on that machine, so Airbyte appears unable to fetch remote connector documentation/definitions �. Has anyone successfully run Airbyte v2 OSS in a true air-gapped setup and resolved this? If yes, what was the workaround — image mirroring only, config change, patching the web app/backend, or a specific version upgrade?

Hello I'm currently building a custom connector for Facebook. I'm aware that Airbyte already offers a default Facebook connector. Has anyone on the team customized it before?

"Production completely blocked by IPC platform bug in source-declarative-manifest:7.21.2. Ticket #18116 - need engineering escalation. Connector works in Builder, fails in production sync. Can anyone from Airbyte team help?"

It is a platform issue and i either need to roll back to prior version 7.20 or have them fix with emergency patch

Hi team — following up on my message from May 13 (https://airbytehq.slack.com/archives/C054V9JFTC6/p1778657050387279). Still no human review on either of my PRs for source-amazon-seller-partner. Bug fix (ready to review): https://github.com/airbytehq/airbyte/pull/77620 Fixes Report Options not being passed to the API for GET_LEDGER_*_VIEW_DATA. Tests added, E2E passing. The bot's /ai-review re-trigger has been broken since May 26 (workflow_dispatch 500 error) — could a maintainer please trigger it manually? Feature (needs direction): https://github.com/airbytehq/airbyte/pull/76434 Adds FbaInboundShipments / FbaInboundShipmentItems streams. Has merge conflicts I can resolve once a maintainer confirms whether to proceed or close in favor of #76436. Thank you.

Hi there. I have submitted an issue with a proposed solution in this PR. I have resolved all the CI checks except one. The Progressive Rollout Gate check is failing with:

Error: GCP_PROD_DB_ACCESS_CREDENTIALS environment variable is not set.

Could a maintainer please re-run the Progressive Rollout Gate from the Airbyte org side (which has access to the secret)? Or are you able to create a pre-release of the connector?

Hi team, I’m seeing unexpected behaviour with Airbyte soft deletes on Databricks Lakehouse. The docs indicate only the _ab_cdc_deleted_at field should be populated, but I’m observing that all fields are also being updated to null. Has anyone run into this before?

Hi team, have a number of mongodb open issues w/ associated PRs approved by the bots. What is the process for getting those across the finish line? https://github.com/airbytehq/airbyte/issues/80160 https://github.com/airbytehq/airbyte/issues/80211 https://github.com/airbytehq/airbyte/issues/80250

Hello! I would love to get this salesforce connector updated merged in: github.com/airbytehq/airbyte/pull/80892, Salesforce is about about to start enforcing refresh token rotation soon (their official date is June 25th), without token rotation support Salesforce oauth'd airbyte connections will start to fail

Hi Airbyte Team 👋 I proposed this PR to make the

destination-gcs

connector much more fast thanks to sockets, aligned with destination-s3 connector 🏎️ Could you help me to review, validate and merge this PR ? 🙏 Thanks a lot !

Hello community, I reported an issue regarding the Connector Builder / Declarative CDK a while ago, and I would heavily appreciate it if someone from the team could take a look, as it severely impacts sync performance. When using

incremental_dependency: true

on parent-child streams, the child stream drops its cursor context after the first sync, forcing the parent stream into an unexpected full refresh loop on subsequent syncs (making them 100x–1000x slower). I just updated the issue with a comprehensive root-cause analysis (assisted by GitHub Copilot and verified against the latest Concurrent CDK source code). I also provided: 1. A functional connector-level workaround using custom components (StateMigration) that anyone can apply today. 2. The exact lines in the CDK core (

substream_partition_router.py

) where the context is being lost. Here is the link to the issue: github.com/airbytehq/airbyte/issues/61567 Could you please help review the proposed architectural fix or provide recommendations to get this merged? Thanks!

Requesting a CDK maintainer to review/sponsor a small Load-CDK fix — it's out of the standard community funnel since it touches the destination/CDK path, so flagging here per the contributing guide. PR: github.com/airbytehq/airbyte/pull/78624 · Issue: github.com/airbytehq/airbyte/issues/81442 (+26/−3, CLA signed, changelog + version bumped, unit tests added) Bug: destination-s3-data-lake with a REST catalog that vends STS-scoped credentials (e.g. Lakekeeper) fails with 403 AccessDenied on s3:ListBucket. IcebergTableCleaner.clearTable() passes the table location without a trailing slash into deletePrefix() → listPrefix() → ListObjectsV2. When the vended session policy scopes s3:prefix to "<table>/*", the un-slashed prefix is denied (and it can also match sibling tables like orders vs orders_archive/). Fix: one line — normalize to a trailing slash before deletePrefix (idempotent; mirrors Iceberg's own FileSystemWalker and Trino's directoryKey()). Validation: built a custom connector image with the patch and confirmed end-to-end against Lakekeeper with STS credential vending — stock connector fails CHECK with the 403; patched connector passes and completes a full sync into Iceberg. The failing integration checks on the PR are the known fork-secrets limitation (empty GCP_GSM_CREDENTIALS). Could a CDK maintainer take a look?

A second small CDK fix for destination-s3-data-lake, again for REST catalogs that enforce multi-level namespaces (Lakekeeper / Polaris). Flagging here sinc it's Load-CDK/destination code — looking for a maintainer to review/sponsor. PR: github.com/airbytehq/airbyte/pull/81451 · Issue: github.com/airbytehq/airbyte/issues/81450 (CDK version + changelog bumped, unit test added) The bug: setting a dotted destination namespace — e.g. operational.lims.inventory.bronze (a multi-level Iceberg namespace) — fails at CHECK/sync:

org.apache.iceberg.exceptions.BadRequestException: Malformed request: Namespace parts cannot contain '.'

TableIdGenerator.tableIdOf builds the id with Namespace.of(namespace), so the whole dotted string becomes a single namespace level containing dots. Iceberg treats . as the level separator and REST catalogs reject it — so the connector can only ever write to one flat namespace. The fix: split on . into levels (single-level names unaffected):

TableIdentifier.of(Namespace.of(*namespace.split(".").toTypedArray()), name)

Matches how Trino, PyIceberg, and the Iceberg REST spec interpret dotted names. Validation: built a custom image with this fix and replicated ~457k rows across 5 nested namespaces (operational.<domain>.bronze) into Lakekeeper, where the stock connector fails with the error above.

Hi team 👋 . Could I get a maintainer review on a source-gong fix? PR #82260 — github.com/airbytehq/airbyte/pull/82260 Fixes #78480 (source-gong 429 / rate-limit failures) What it does:

callTranscripts

currently sends one POST /calls/transcript request per call (SubstreamPartitionRouter with a single callId), so backfills blow past Gong’s ~3 req/s rate limit and ~10k/day quota. The PR wraps the existing router in a GroupingPartitionRouter (group_size: 100) so one request covers up to 100 calls — ~99% fewer requests. Manifest-only change + a unit test; preserves the private-call filter and started cursor, no schema change/reset. Validated against a live Gong tenant: array callIds accepted, 100/request → 200, cursor pagination works alongside callIds. Status: CLA signed ✅, all fork-accessible checks green. The only red checks are the fork “No Creds” connector test and the progressive-rollout gate — both need maintainer-side action (/run-connector-tests + rollout bypass). Would appreciate someone running the credentialed suite and taking a look 🙏

Hello, I have just opened an urgent pull request on the airbyte Goldcast connector to fix the connector following some un-versioned changes done on their API. Could we please get a review as soon as possible? • Issue: github.com/airbytehq/airbyte/issues/83236 • PR: github.com/airbytehq/airbyte/pull/83237 Thanks 🙏

Hi everyone, I’ve submitted a feature request to Airbyte for configuring CPU and memory resources per connection directly from the UI. Airbyte already supports connection-specific resource requirements internally, but today this typically requires changing the internal database manually. In environments with many connections, this makes resource management harder than it should be. The main use case is that not every connection has the same workload. For example, a lightweight CDC connection may only need a small amount of CPU and memory, while another connection doing a large initial sync or heavy batch load may need much more. Being able to configure these values per connection from the UI would help avoid both over-provisioning and under-provisioning. If this would be useful in your environment as well, please consider upvoting the feature request. More upvotes can help show the Airbyte team that there is broader demand for this capability. github.com/airbytehq/airbyte/discussions/85061

Hi, everyone. I have identified an issue with MySQL 8.4 tagged GTIDs support and CDC. Here is my contribution github.com/airbytehq/airbyte/pull/85855

Hi all, I've submitted two PRs with fixes that are ready for review but need a maintainer to trigger /run-connector-tests. 1. Issue: github.com/airbytehq/airbyte/issues/85815 PR: github.com/airbytehq/airbyte/pull/85820 (source-tiktok-marketing — adds missing Website and Shop conversion metrics to daily report streams) 2. Issue: github.com/airbytehq/airbyte/issues/86324 PR: github.com/airbytehq/airbyte/pull/86358 (source-recharge — fixes a bug where the events stream crashes the entire sync for any connection with a start_date older than 7 days) Both PRs are linked to their issues, CLA is signed, and all fork-runnable checks are green. The remaining red checks are the expected fork-secret limitation. Would someone be able to run /run-connector-tests and take a look? These are blocking a client migration on our side, so any help would be appreciated. Thanks!

Hi team 👋 please could you consider to review and merge this small

LinkedIn Ads

PR fix on

videos

stream ? github.com/airbytehq/airbyte/pull/86321 Thanks a lot

Hi! Could someone please triage #86962⁠ and take a look at PR #86966⁠ for

source-shopify

? It fixes silent data loss when a failed Shopify bulk operation returns neither

url

nor

partialDataUrl

. The PR includes a regression test; unit tests, integration tests, lint, formatting, and image build pass. The remaining CI failures require Airbyte’s internal credentials Thanks a lot

Your harness uses MySQL EOL endoflife.date/mysql You should update harness to support 8.0.46 and 8.4.x and 9.x

Evidence Harness: Local source-mysql-e2e-cdc-tests harness (MySQL 8.0.46, binlog+GTID, CDC read with ssl_mode=verify_ca + client cert/key against a REQUIRE X509 replication user), control vs target with --reset=fixture; plus Ops regression suite (GSM creds) for the baseline.

github.com/airbytehq/airbyte/pull/77840

Hi team, made a contribution to the ashby source to add incremental syncs on applications and application_history. Would love it if someone could take a look 🙂 github.com/airbytehq/airbyte/pull/87597

Hi folks, could someone take a look at this issue/PR that allows the snowflake destination connector to connect snowflake through a proxy at an arbitrary domain, it's super short https://github.com/airbytehq/airbyte/issues/87719

Hi everyone, We are currently planning a major upgrade of our self-hosted Airbyte instance, migrating from version

0.59.1

to

2.3.0

. We rely heavily on the Declarative CDK and have dozens of custom connectors. Because of this scale, recreating each of them manually from scratch via the Connector Builder UI is completely unfeasible for our team. Initially, we tried the straightforward approach: exporting the YAML manifest from the obsolete version and importing it directly into the new version's Connector Builder. However, this approach throws validation errors and fails to save the draft. Since a simple UI import doesn't work for these legacy manifests, we are looking for suggestions on how to best handle this transition. What is the officially recommended approach to migrate and deploy these legacy YAML manifests into v2.3.0? Any guidance or documentation you could share would be hugely appreciated! Thanks in advance.

Hey folks, could someone take a look at this issue/PR for an API connector? github.com/airbytehq/airbyte/pull/86327 I think I've got it ready for review, but very open to feedback

Hi, looking for a maintainer review on a small

source-pylon

community PR: #87522 Adds parent IDs (

issue_id

/

knowledge_base_id

) to two child streams via

AddFields

. Rebased onto current master (

0.0.29

); connector CI is green. Progressive Rollout Gate is failing on missing

GCP_PROD_DB_ACCESS_CREDENTIALS

in CI (empty secret), not on the change itself. Thanks!

Same here, need a maintainer review, github.com/airbytehq/airbyte/issues/33039 this issue should be resolved by github.com/airbytehq/airbyte/pull/90145 this PR allowing for OIDC (IAM role) execution of destination-redshift as well as allowing a key for fallback.