neoclide/coc.nvim · error

Request failed using proxy ${proxy.host}: ${err.message}

Error message

Request failed using proxy ${proxy.host}: ${err.message}

What it means

fetch() wraps any request failure: if the request used a proxy agent (opts.agent.proxy set), the original error is replaced with 'Request failed using proxy <proxy.host>: <original message>' after logging the underlying error. This contextualizes network failures as proxy-related, since a misconfigured or unreachable proxy is the likely cause when a proxy agent is active.

Source

Thrown at src/model/fetch.ts:330

/**
 * Send request to server for response, supports:
 *
 * - Send json data and parse json response.
 * - Throw error for failed response statusCode.
 * - Timeout support (no timeout by default).
 * - Send buffer (as data) and receive data (as response).
 * - Proxy support from user configuration & environment.
 * - Redirect support, limited to 3.
 * - Support of gzip & deflate response content.
 */
export default function fetch(urlInput: string | URL, options: FetchOptions = {}, token?: CancellationToken): Promise<ResponseResult> {
  let url = toURL(urlInput)
  let opts = resolveRequestOptions(url, options)
  return request(url, options.data, opts, token).catch(err => {
    logger.error(`Fetch error for ${url}:`, opts, err)
    if (opts.agent && opts.agent.proxy) {
      let { proxy } = opts.agent
      throw new Error(`Request failed using proxy ${proxy.host}: ${err.message}`)
    } else {
      throw err
    }
  })
}

View on GitHub (pinned to 50e974d969)

Solutions

1. Check the underlying err.message after the colon — it names the real cause (ECONNREFUSED, ETIMEDOUT, 407, certificate error).
2. Verify the proxy host/port in configuration (e.g. http.proxy) and that the proxy is reachable: curl -x http://host:port https://example.com.
3. If no proxy is needed, clear the proxy setting so the request goes direct and surfaces the raw error.
4. For HTTPS through the proxy, configure proxyCA (options.proxyCA) if the proxy re-signs TLS.
5. Retry after network/proxy recovery; transient proxy outages are common.

Example fix

// before (config)
{ "http.proxy": "http://127.0.0.1:8080" } // proxy down
// after
{ "http.proxy": "" } // or point at the correct proxy
// then retry the fetch

const net = require('net')
async function proxyReachable(host, port) {
  return new Promise(res => {
    const s = net.connect({ host, port, timeout: 3000 }, () => { s.destroy(); res(true) })
    s.on('error', () => res(false)); s.on('timeout', () => { s.destroy(); res(false) })
  })
}
// only call fetch if await proxyReachable(proxyHost, proxyPort)

try {
  return await fetch(url)
} catch (e) {
  const m = /^Request failed using proxy (.+?): (.*)$/.exec(e.message)
  if (m) {
    logger.warn(`proxy ${m[1]} failed: ${m[2]}; retrying without proxy`)
    return fetchWithoutProxy(url)
  }
  throw e
}

When it happens

Trigger: A request through coc's configured HTTP proxy fails — proxy host unreachable, proxy rejects CONNECT, wrong credentials, proxy timeout, or TLS failure against the proxy — producing e.g. 'Request failed using proxy 127.0.0.1:8080: connect ECONNREFUSED 127.0.0.1:8080'.

Common situations: Corporate proxies requiring authentication not configured in coc settings; VPN toggled off so the proxy address is unreachable; httpProxy pointing at a dead port; proxy CA not trusted (proxyCA not set).

Related errors

• Unsupported charset: ${encoding}
• Not valid protocol with ${urlInput}, should be http: or http
• maxResponseSize must be a positive finite number

AI-assisted analysis of neoclide/coc.nvim@50e974d969 (2026-08-31). Data as JSON: /api/errors/e4b1b28424f16287. Report an issue: GitHub.