Troubleshoot Handover
Resolve Handover access, Drive validation, processing, integrity, checklist, export, and recovery problems.
Use the visible run, file, and checklist states to identify the narrowest safe recovery. Do not work around an access error by sharing a broader Drive folder or granting a stronger Drive role.
You cannot find or open Handover
Handover is absent from shared navigation
This is expected during the gated launch. Handover is hard-hidden from the Venture sidebar, app switcher, and command palette. Open an approved direct link on handover.infra.one.
The host root says No organization selected
Use the mandatory organization selector. It cannot be dismissed at the root. The selector lists organizations based on membership and rollout state; it does not confirm that you have management permission.
If the organization is missing, confirm that you are a member and that fund_handovers is enabled for the organization or your account. If it appears but the destination denies access, confirm that you are an organization admin with manage_fund_handovers.
A deep link sends you to Venture
For an approved canonical Handover deep link, complete authentication, including profile completion, email confirmation, or pending two-factor authentication when requested. Venture should return you to the exact Handover host, path, and query after sign-in.
Venture can briefly show Finishing sign-in… with You’ll be redirected automatically. If automatic navigation does not continue, under Not redirected?, select Continue to your destination.
If an old Venture Handover link leaves you on Venture after sign-in, open the approved canonical handover.infra.one link instead. Do not repeat sign-in. For an already authenticated user, an old /app/orgs/:organization-slug/handovers link redirects permanently to the canonical Handover host. The final Handover route still applies rollout, membership, permission, and run-ownership checks.
You cannot start a run
| Message | Cause and recovery |
|---|---|
| That does not look like a Google Drive folder link. | Paste a supported Drive folder URL, scheme-less folder URL, open?id= URL, or bare folder ID. Remove surrounding prose. |
| The folder is not shared with the service account yet — share it with <email> as viewer, then try again. | Share the submitted folder with the exact address displayed under 1. Share the folder, choose Viewer, wait for Drive to apply the permission, and retry. |
| The folder is not readable by the service account. | Recheck the folder ID, displayed account, and Viewer permission. Drive did not return a more specific validation reason. |
| That link points to a single file — paste the link of the folder that contains the handover. | Submit the containing folder. A root shortcut to a file is also rejected; file shortcuts are supported only when discovered inside a walked folder. |
| A run for this folder is already in progress — open it from the handovers list instead. | Open the active run from Handovers. After it reaches a terminal state, use Re-run when another pass is needed. |
| Google Drive is rate-limiting requests right now — try again in a minute. | Wait at least a minute before validating the same folder again. |
| The handover service account is not configured for this environment. | Stop and ask the environment owner to configure the service account. Do not substitute another address. |
| Something went wrong while checking the folder — try again. | Retry once. If it repeats, note the time and exact message, then share the organization and folder ID only in an approved private support channel. Never copy private file names into support. |
When configuration is absent, the new-run page can also show The handover service account is not configured for this environment — runs cannot be created.
A run is slow, stalled, or failed
The status jumps over Evaluating checklist…
This is current behavior, not evidence that evaluation was skipped.
Evaluating checklist… is a declared run label, but the current finalizer never writes the evaluating state; a normal run can move directly from Fetching files… to Completed.
Drive is rate-limiting background work
Discovery automatically retries from its saved folder frontier. Retryable failures have a five-failure budget; Drive rate limits instead snooze the job for 60 seconds without consuming that budget. File fetching has the same five-failure budget and snoozes for 90 seconds on a Drive rate limit. Repeated rate limits can therefore produce more than five executions. Let the automatic retry finish instead of starting a duplicate run.
If discovery exhausts its attempts, the run becomes Failed. If one file exhausts its attempts, that item becomes Failed, but the run can still become Completed.
Drive access was removed during processing
If access is revoked while Handover walks the folder, the run can fail. Restore Viewer access for the displayed service account, wait for the run to settle, then use Re-run. Loss of access during file fetching can fail the affected item without failing the complete run.
The service-account configuration changed during processing
Missing configuration during folder discovery fails the run. During file fetching, it can fail the affected item. Ask the environment owner to restore configuration, then re-run only after the current run settles.
Files are missing, skipped, or unusable
The file table shows Skipped or Failed, but it does not expose the internal skip or failure reason. Use the source state and reported size as context; support can confirm the recorded reason when needed.
| Symptom | Explanation and next step |
|---|---|
| No files discovered yet. after a terminal state | Clear any statistic or checklist drill-down, search, and Entity Browser filters, then return to the root folder. If the empty state remains and Files found is zero, confirm the submitted root contains files reachable by the service account. Re-share the intended root, then re-run. |
| Skipped, with a reported ordinary-file size above the cap | A Drive-reported size over 250,000,000 bytes is rejected before download. Ask the sender for a supported smaller delivery or split document. If Drive omitted or understated the size, the storage backstop can instead leave the oversized item Failed. |
| Skipped, with a reported Google-native size above the export precheck | A Google-native file reported more than 10,000,000 bytes before export. Ask the sender for a smaller or pre-exported copy. |
| A pre-existing queued or fetching item becomes Skipped after a later discovery walk | The walk did not rediscover that leftover item. Restore the source and re-run if it is still required. A disappearance during active download can instead exhaust retries and leave the item Failed. |
| Empty file | The zero-byte source otherwise passed storage safety rules, was stored, and was classified as Note (filename is the message). It does not satisfy or degrade a checklist requirement, so the underlying requirement remains Missing unless another healthy file matches. Ask for a nonempty replacement. |
| Corrupt at source | Replace the source with a readable copy, then re-run. |
| Password-protected | Ask for an approved readable copy. Do not put a password in a file name, ticket, or documentation. |
| A PDF is implicated in a Degraded checklist row while Integrity problems is zero | The internal first-three-page text assessment can treat a scan as unusable without increasing the counter. The file table and File Viewer do not display that assessment, so use the stored file’s View action or Download, or ask support to confirm it. Supply a text-searchable PDF or another healthy matching document when checklist coverage is required. |
| Blocked executable or script content | Deliver the required information in a supported, reviewable format. Do not rename an executable to bypass the block. |
Google Docs, Sheets, Slides, and Drawings are exported to PDF. File shortcuts found inside the tree are fetched through their targets, and folder shortcuts are followed with cycle protection. ZIP and macro-enabled Office files can be stored, but content inside nested archives is not separately classified or matched to checklist requirements.
The file table pages 50 items at a time. Folder navigation exposes at most 1,000 immediate folders per level and omits paths longer than 1,024 bytes. Those navigation boundaries do not establish that a source file was absent from the inventory; search the file table before concluding it was not discovered.
Checklist results look wrong
A row is Present even though one match is unusable
One healthy matching file makes the row Present, even when another match is corrupt, password-protected, or a scanned PDF without text. Select View N files and review every match.
A row is Degraded but Integrity problems is zero
Integrity problems counts corrupt and password-protected files only. Handover’s internal first-three-page text assessment can make a scanned PDF’s row Degraded without increasing that counter, but neither the current file table nor File Viewer displays the assessment. Select View N files, then use the stored file’s View action or Download, or ask support to confirm it. Empty files that otherwise pass storage safety rules are also excluded from the counter, but they are classified as Note (filename is the message) and do not create a degraded checklist row.
More scopes appear than expected
Checklist expansion is evidence-driven. Investor rows come from investor-specific document categories; quarter rows span continuously from the earliest through latest observed reporting quarter; year rows use observed reporting years; and drawdown rows run from 1 through the highest inferred number.
Evaluation covers the entire run. If one Drive root contains several funds or vehicles, their evidence can combine. Start separate runs with one fund or vehicle per root when fund-specific completeness is required.
Mark needed or Ignore did not change the request list
That is expected. Manual file review and checklist matching are independent. Mark needed does not add a checklist gap, and Ignore does not suppress one.
The request-list export is empty or stale
Handover renders request-list downloads in every run state. Before Completed, an initial run exports no checklist rows and a re-run continues to export the previous checklist snapshot, which can also have no rows, until final evaluation atomically replaces it. Wait for Completed before interpreting or downloading either export.
After Completed:
- CSV must contain Requirement, Scope, Severity, and Status.
- Markdown shows Required or Nice to have only when that severity has rows, and marks an unusable match present but unusable, please re-send.
- Both formats include Missing and Degraded rows only. A Present row is deliberately absent.
- A missing or unknown
formatquery selects Markdown; useformat=csvexplicitly for CSV.
Choose a safe recovery action
- Use Cancel only for a nonterminal run you intend to stop. Queued and executing jobs are canceled, but files keep the states they reached.
- Use Re-run after a terminal state when Drive content or access changed. Rediscovered regular files and direct targets are reclassified without another download when unchanged; eligible entries detected as changed, failed, or skipped are fetched again. A known shortcut-only entry is not redispatched on re-run, so use a new run or add the direct target when it must be refreshed.
- Use Delete only when the run and its stored copies should be removed. Handover cancels active work if needed, deletes the database records, and schedules stored blobs for asynchronous cleanup.
A settled file whose Drive source disappears remains historical evidence across re-runs. Re-running does not purge that terminal item; use a separate new run when you need a current-only inventory. If an export or cleanup remains inconsistent after the appropriate terminal action, escalate rather than repeatedly creating or deleting runs.
Information to collect for support
Record the terminal run status, time of the failure, exact visible message, affected file status, and whether Drive Viewer access is still present. Share the organization slug, run ID, and folder ID only in an approved private support channel. Never include service credentials, access tokens, customer documents, file names, passwords, or a broader Drive share.