stablyai/orca · error · Error

Remote connection dropped. Click Reconnect on the SSH target

Error message

Remote connection dropped. Click Reconnect on the SSH target before retrying.

What it means

Thrown by the 'git:status' handler (src/main/ipc/filesystem.ts:1182) when args.connectionId is set but getSshGitProvider(connectionId) returns undefined. The SSH git provider map (ssh-git-dispatch.ts) is keyed by connectionId and a provider is removed via unregisterSshGitProvider when the SSH connection drops. The accompanying generation counter increments on each register/unregister so callers can detect a stale connection.

Source

Thrown at src/main/ipc/filesystem.ts:1182

      }
    ): Promise<GitStatusResult> => {
      const controller = gitStatusCancellations.begin(event, args.requestToken)
      const options = {
        includeIgnored: args.includeIgnored ?? false,
        ...(args.reuseLineStats === true ? { reuseLineStats: true } : {}),
        ...(args.branchLineTotalMergeBase === undefined
          ? {}
          : { branchLineTotalMergeBase: args.branchLineTotalMergeBase }),
        ...(args.bypassEffectiveUpstreamNegativeCache === true
          ? { bypassEffectiveUpstreamNegativeCache: true }
          : {}),
        ...(controller ? { signal: controller.signal } : {})
      }
      try {
        if (args.connectionId) {
          const provider = getSshGitProvider(args.connectionId)
          if (!provider) {
            throw new Error(SSH_GIT_PROVIDER_UNAVAILABLE_MESSAGE)
          }
          // Why: await keeps the cancellation token registered until the remote request settles (an early finally would free it).
          return await provider.getStatus(args.worktreePath, options)
        }
        const worktreePath = await resolveRegisteredWorktreePath(args.worktreePath, store)
        // Why: one registered-worktree lookup feeds both — status polls this
        // handler, and the scan walks every repo's worktree meta.
        const repo = getLocalRepoForRegisteredWorktree(store, args.worktreePath, worktreePath)
        const gitOptions = getLocalGitOptionsForRepo(store, repo)
        const sharedLinkPaths = repo ? getWorktreeSharedLinkPaths(repo) : []
        return await getStatus(worktreePath, {
          ...options,
          ...gitOptions,
          ...(sharedLinkPaths.length > 0 ? { sharedLinkPaths } : {})
        })
      } finally {
        gitStatusCancellations.finish(event, args.requestToken, controller)
      }

View on GitHub (pinned to 1136503c6a)

Solutions

1. Trigger the app's 'Reconnect' on the SSH target to re-register an SSH git provider.
2. After reconnect, refresh the connectionId in the renderer (reconnect yields a new id/generation).
3. Retry the git:status call once the reconnect is confirmed.
4. If it persists, the SSH target is unreachable: verify network, credentials, and that the relay process is up.

Example fix

// before: invoke status with a stale connectionId
await invoke('git:status', { worktreePath, connectionId })

// after: reconnect first, then retry with the fresh id
if (!await ensureSshConnected(connectionId)) {
  connectionId = await reconnectSshTarget(targetId)
}
await invoke('git:status', { worktreePath, connectionId })

// Treat a missing provider as a reconnect signal; check connection health first.
async function ensureProvider(connectionId: string): Promise<string> {
  if (await isSshConnected(connectionId)) return connectionId
  return reconnectSshTarget(connectionId)
}

const UNAVAILABLE = /Remote connection dropped|No git provider for connection/
try {
  return await invoke('git:status', { worktreePath, connectionId })
} catch (e) {
  if (e instanceof Error && UNAVAILABLE.test(e.message)) {
    connectionId = await reconnectSshTarget(targetId)
    return invoke('git:status', { worktreePath, connectionId }) // one retry
  }
  throw e
}

When it happens

Trigger: Invoking ipcRenderer.invoke('git:status', { worktreePath, connectionId }) while the SSH connection for that connectionId has disconnected, been torn down, or was never registered. Also after a reconnect that produced a new provider/connectionId while the renderer still holds the stale id.

Common situations: SSH TCP session dropped by network sleep/resume; remote relay restarted; rapid branch/target switching racing with teardown; renderer holding a connectionId from before a reconnect.

Related errors

• No git provider for connection "${args.connectionId}"
• Remote connection dropped. Click Reconnect on the SSH target
• Clone failed: ${message}
• Translation request failed with status ${response.status}
• Unable to load source control

AI-assisted analysis of stablyai/orca@1136503c6a (2026-08-12). Data as JSON: /api/errors/43b860a91d4a3c8c. Report an issue: GitHub.