QuantumNous/new-api · error · Error

HTTP ${response.status}: ${response.statusText}

Error message

HTTP ${response.status}: ${response.statusText}

What it means

Thrown by the Ollama model-pull dialog when the streaming response to its fetch is not ok or has no body. The request posts channel_id and model_name to a SSE endpoint with Accept: text/event-stream; any non-2xx status (or missing body) aborts before the reader loop starts, producing this HTTP status string.

Source

Thrown at web/src/features/channels/components/dialogs/ollama-models-dialog.tsx:264

    try {
      const authHeaders = await getFreshAuthHeaders()
      const response = await fetch('/api/channel/ollama/pull/stream', {
        method: 'POST',
        credentials: 'include',
        headers: {
          ...authHeaders,
          Accept: 'text/event-stream',
        },
        body: JSON.stringify({
          channel_id: channelId,
          model_name: pullName.trim(),
        }),
        signal: controller.signal,
      })

      if (!response.ok || !response.body) {
        throw new Error(`HTTP ${response.status}: ${response.statusText}`)
      }

      const reader = response.body.getReader()
      const decoder = new TextDecoder()
      let buffer = ''

      while (true) {
        const { done, value } = await reader.read()
        if (done) break

        buffer += decoder.decode(value, { stream: true })
        const lines = buffer.split('\n')
        buffer = lines.pop() || ''

        for (const line of lines) {
          if (!line.startsWith('data: ')) continue
          const eventData = line.slice(6)
          if (!eventData) continue

View on GitHub (pinned to e2c7aa7b10)

Solutions

1. Match the status: 401/403 → re-login/fix channel key; 404 → wrong endpoint or model name; 502/504 → proxy timeout or Ollama host down
2. Verify the Ollama server in the channel settings is reachable from the backend (curl the Ollama /api/tags from the backend host)
3. Increase proxy read timeout / disable buffering for the SSE route if pulls are long
4. Retry with the exact model tag from `ollama list` on the target host

Example fix

// before
if (!response.ok || !response.body) {
  throw new Error(`HTTP ${response.status}: ${response.statusText}`)
}
// after - include the response body message when present
if (!response.ok || !response.body) {
  let detail = `${response.status}: ${response.statusText}`
  try {
    const body = await response.json()
    if (body?.message) detail = `${detail} - ${body.message}`
  } catch { /* body not JSON */ }
  throw new Error(detail)
}

const name = pullName.trim()
if (!name) { toast.error(t('Model name is required')); return }
if (!channelId) { toast.error(t('Channel is not selected')); return }

try {
  const response = await fetch(url, { headers: { ...authHeaders, Accept: 'text/event-stream' }, body: JSON.stringify({ channel_id: channelId, model_name: name }), signal: controller.signal })
  if (!response.ok || !response.body) {
    let detail = `HTTP ${response.status}: ${response.statusText}`
    try { const b = await response.json(); if (b?.message) detail += ` - ${b.message}` } catch {}
    throw new Error(detail)
  }
  // reader loop
} catch (error) {
  toast.error(error instanceof Error ? error.message : 'Pull failed')
}

When it happens

Trigger: Submitting a model pull where the backend returns 4xx/5xx: unknown channel_id, model_name blank after trim, upstream Ollama server unreachable, model not found in the registry, or the long-lived request being killed by a proxy (502/504).

Common situations: Ollama host in the channel config wrong or down; reverse proxy (nginx) buffering/timing out SSE; model name typo so the registry returns 404; admin session expired giving 401; channel key invalid giving 403.

Related errors

• Failed to fetch status

AI-assisted analysis of QuantumNous/new-api@e2c7aa7b10 (2026-08-15). Data as JSON: /api/errors/53a38350a7b6b422. Report an issue: GitHub.