TABLE OF CONTENTS
Source updates and reports are not instant. Connectors run on schedules, process work in batches, and pass changes through separate services. Reporting adds another data pipeline and groups results into time periods. A successful change at the source may therefore take more than one step before every Stack Internal screen shows it.
Source sync timing
Each connector keeps a checkpoint so it can continue from the last safe point. A checkpoint moves only after the related content is stored successfully. After storage, other services extract searchable knowledge and update their own views. This separation prevents one failure from losing the source position, but it also makes updates eventually consistent.
| Source | Current default schedule | What can extend the delay |
|---|---|---|
| Stack Internal Community | Every 10 minutes | Questions, answers, and articles are processed in bounded batches. A backlog can need later runs. |
| Slack | Every 2 hours | Each selected channel admits up to 50 top-level messages per run. Large threads, attachments, throttling, or many channels can take longer. |
| Google Drive | Once each day | The connector pages through Drive and then fetches each native Google Doc, its comments, and its permissions. API errors can defer unfinished work. |
| Source | Supported searchable content | Current processing limit or boundary |
|---|---|---|
| Google Drive | Native Google Docs in configured drives. The connector also reads document comments and sharing permissions. | Sheets, Slides, PDFs, Microsoft Office files, images, shortcuts, and other Drive file types are skipped. No customer file-size or file-count ceiling is stated in the current implementation. |
| Slack | Messages and thread replies from selected channels. | Up to 50 top-level messages per selected channel per run. Replies for those messages do not count against the 50. If no start date is set, the default history window is three months. |
| Stack Internal Community | Questions, answers, and articles. | The deployed default is 100 questions, 100 answers, and 100 articles per run, using a page size of 15. The checkpoint continues later. |
| Schedules are defaults, not a promise for one exact clock time The service spreads tenant jobs across the schedule to avoid a traffic spike, and an intentionally customized schedule can be preserved. Enabling a connector is normally seen by its next scheduled execution. Processing after that execution has no published end-to-end time guarantee in the sources reviewed. |
|---|
When to contact Support about sync timing
First wait through the source’s next expected run and allow extra processing time. If the item still does not appear, check the source status. “Needs attention” means the connector found a problem that requires an administrator. A temporary rate limit does not normally show as “Needs attention”; the connector leaves the source connected so a later run can retry.
Analytics and reporting delay
Reporting follows a longer path than operational source sync. Product events are captured to durable storage, loaded into a data warehouse, shaped into reporting tables, and then read by the Reporting service. Event capture itself adds minutes of delay. Later modelling and publication add more time.
Stack Internal reporting is typically about one day behind. A daily process loads report data into the reporting database.
Supported content and processing limits
| Source | Supported searchable content | Current processing limit or boundary |
|---|---|---|
| Google Drive | Native Google Docs in configured drives. The connector also reads document comments and sharing permissions. | Sheets, Slides, PDFs, Microsoft Office files, images, shortcuts, and other Drive file types are skipped. No customer file-size or file-count ceiling is stated in the current implementation. |
| Slack | Messages and thread replies from selected channels. | Up to 50 top-level messages per selected channel per run. Replies for those messages do not count against the 50. If no start date is set, the default history window is three months. |
| Stack Internal Community | Questions, answers, and articles. | The deployed default is 100 questions, 100 answers, and 100 articles per run, using a page size of 15. The checkpoint continues later. |
Rate limits, retries, and quotas
Upstream services can slow or reject connector calls. Stack Internal handles the cases it recognizes and protects checkpoints so unfinished content can be retried.
| Situation | What the product does | What you should do |
|---|---|---|
| Slack API throttling | The connector limits itself to 100 calls per minute, queues up to 50 calls, honors Slack’s Retry-After value, and retries a throttled or temporary request up to three times. | Usually wait for the next connector run. Escalate if the same channel remains behind after another expected run. |
| Repeated temporary failures | A circuit breaker pauses Slack calls for 30 seconds after a high failure rate. The affected checkpoint is not advanced. | Check Slack service health and wait. Do not reconnect unless the source shows an administrator-action error. |
| Invalid or revoked credentials | The source moves to an error state that the Admin UI labels “Needs attention.” | Reconnect the source or ask a workspace administrator to restore access. |
| Stack Internal API rate | API calls are limited on a monthly basis and defined by your purchased credit band. Free plans are limited to 1,000 credits a month. | Contact your sales representative to purchase additional credits. |
Support diagnostics
Send enough detail to locate the run or request without sending source content, tokens, secrets, or personal data that is not needed for the investigation.
Include these details
- Workspace name and workspace slug.
- Affected source: Google Drive, Slack, Stack Internal Community, SSO, API, MCP, or Analytics.
- The affected user’s work email, when the problem is user-specific.
- Source item URL or stable ID, channel ID, drive ID, or report name. Do not copy the document or message body unless Support asks for it through an approved channel.
- When the source item changed, including time zone, and when you last checked Stack Internal.
- The last time the same action worked, if known.
- Exact error text, HTTP status, and any request or correlation ID shown to you.
- A redacted screenshot that shows the state and time without exposing tokens or private content.
- Whether the source page shows Connected, Running, Disconnected, or Needs attention.