The process memory usage is rising in AJAX calls.
Reported by
k078...@gmail.com,
Jan 2
|
|||||
Issue description
Chrome Version: Chromium 72.0.3618.0 (Headless)
OS: Ubuntu 18.04
macOS 10.14
<b>See <https://www.chromium.org/for-testers/providing-memory-details> to</b>
provide memory usage snapshots.
URL (if applicable) where the memory bloat occurred:
Can you reproduce this memory bloat?
What steps will reproduce this memory bloat (or if it's not reproducible,
what were you doing until then)?
(1) Make an Ajax request every second.
```html
<!doctype html>
<html>
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, maximum-scale=1, user-scalable=no">
<title>Ajax Test</title>
<script src="https://ajax.googleapis.com/ajax/libs/jquery/2.1.4/jquery.min.js"></script>
</head>
<body>
<div></div>
<script>
sendReq();
function sendReq() {
$.ajax({
type: "POST",
url: "http://127.0.0.1:3001/post",
success: function (msg) {
var timeout = setTimeout(function() {
sendReq();
clearTimeout(timeout);
}, 1000);
}
});
};
</script>
</body>
```
(2) Take a coffee and a few hours.
*Please note that issues filed with no information filled in above
will be marked as WontFix*
,
Jan 3
This issue looks similar to issue 886569 hence adding luoe@chromium.org to cc list for further inputs. Thanks !
,
Jan 7
Hmm, if Headless has an equivalent of chrome://crashes, anything in it?
,
Jan 7
c#2: the other issue 886569 pertains to sites that produce lots of console logs and DevTools is open. This issue seems unrelated to DevTools.
,
Jan 18
(4 days ago)
Confirmed. I created a simpler repro that doesn't involve jquery, and also goes much faster.
```html
<!doctype html>
<script>
sendReq();
function sendReq() {
const xhr = new XMLHttpRequest();
xhr.open("POST", "http://127.0.0.1/cgi-bin/post");
xhr.onreadystatechange = () => {
if (xhr.readyState === 4 && xhr.status === 200) {
var timeout = setTimeout(() => {
sendReq();
clearTimeout(timeout);
}, 10);
}
};
xhr.send("data");
}
</script>
```
Initial analysis using about:tracing appears to show that the leak is in Blink rather than V8.
Even with the increased speed of my repro, the leak is very slow, so the priority to fix this is low.
,
Jan 18
(4 days ago)
Do you know if this leak retains something browser/network-service side? We have some cases where there was an unreasonable number of URLLoaders around...
,
Jan 21
(2 days ago)
#6 I didn't run with browser tracing enabled, but it looks like browser memory usage is stable.
,
Jan 21
(2 days ago)
My comment #8 was mistaken so I have deleted it. I was actually running the original jQuery repro over the weekend. Memory usage stabilised eventually. There's probably nothing to do here. I'm going to run one more test to be on the safe side.
,
Today
(21 hours ago)
I did an overnight run of my repro in #5, and again the memory usage levelled out after a while. I don't think we have a memory leak. |
|||||
►
Sign in to add a comment |
|||||
Comment 1 by swarnasree.mukkala@chromium.org
, Jan 3