HttpClient的“Going to buffer response body of large or unknown size. Using getResponseBodyAsStream instead is recommended.”警告释疑
Posted in java on 八月 21st, 2010 by kafka0102
使用HttpClient，总是报出“Going to buffer response body of large or unknown size. Using getResponseBodyAsStream instead is recommended.”的WARN日志，定位到HttpClient的源码如下：
public byte[] getResponseBody() throws IOException { if (this.responseBody == null) { InputStream instream = getResponseBodyAsStream(); if (instream != null) { long contentLength = getResponseContentLength(); if (contentLength > Integer.MAX_VALUE) { //guard below cast from overflow throw new IOException("Content too large to be buffered: "+ contentLength +" bytes"); } int limit = getParams().getIntParameter(HttpMethodParams.BUFFER_WARN_TRIGGER_LIMIT, 1024*1024); if ((contentLength == -1) || (contentLength > limit)) { LOG.warn("Going to buffer response body of large or unknown size. " +"Using getResponseBodyAsStream instead is recommended."); } LOG.debug("Buffering response body"); ByteArrayOutputStream outstream = new ByteArrayOutputStream( contentLength > 0 ? (int) contentLength : DEFAULT_INITIAL_BUFFER_SIZE); byte[] buffer = new byte[4096]; int len; while ((len = instream.read(buffer)) > 0) { outstream.write(buffer, 0, len); } outstream.close(); setResponseStream(null); this.responseBody = outstream.toByteArray(); } } return this.responseBody; }
报WARN的条件是 ((contentLength == -1) || (contentLength > limit))，也就是说，或者是返回的HTTP头没有指定contentLength，或者是contentLength大于上限（默认是1M）。如果能确定返回结果的大小对程序没有显著影响，这个WARN就可以忽略，可以在日志的配置中把HttpClient的日志级别调到ERROR，不让它报出来。
当然，这个警告也是有意义的，HttpClient建议使用InputStream getResponseBodyAsStream()代替byte[] getResponseBody()。对于返回结果很大或无法预知的情况，就需要使用InputStream getResponseBodyAsStream()，避免byte[] getResponseBody()可能带来的内存的耗尽问题。
=============================== 华丽的终止符 ================================
相关日志
One Response
留下评论
十月 25th, 2010 at 10:12 上午
[...] 2010年08月21日 — HttpClient的“Going to buffer response body of large or unknown size. Using getResponseBodyAsStrea… [...]