# \[Need Help\] Strange behavior of WebSocket checks
**URL:** https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080
**Category:** Gatling (Open-Source)
**Created:** [August 18, 2022, 1:26pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080 "2022-08-18T13:26:38Z")
**Posts on this page:** 9
**Page:** 1
### Author:  [@nobody](https://community.gatling.io/u/nobody)
#### Post date: [August 18, 2022, 1:26pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/1 "2022-08-18T13:26:38Z")
Hi,
I’m testing a websocket service, the scenario should be:
- Each client keeps the connection open and sends 1 message with a random unique sequence number every 10 seconds.
- Server returns the seq in response body.
- Client should only accept the message with the same seq.
I also want to print the seq for failed requests.
Here’s my code:
```java
public class WebsocketTest extends Simulation {
private static final Logger log = LoggerFactory.getLogger(WebsocketTest.class);
// ...
HttpProtocolBuilder httpConf = http
.wsBaseUrl(baseUrl)
.wsAutoReplySocketIo4();
ScenarioBuilder scn = scenario("WebSocket")
.exec(ws("Connect WS").connect(path))
.exitHereIfFailed()
.repeat(loop).on(
pause(sendIntervalMin, sendIntervalMax)
.exec(session -> {
session.markAsSucceeded();
String nanoTime = String.valueOf(System.nanoTime());
return session.set("seq", System.currentTimeMillis() + nanoTime.substring(nanoTime.length() - 5));
})
.exec(ws("Send text")
.sendText(data)
.await(responseTimeout).on(
ws.checkTextMessage("Check response code")
.matching(
jsonPath(matchJsonPath).ofLong().isEL("#{seq}").saveAs("returnSeq")
)
.check(
jsonPath(checkJsonPath).ofInt().is(returnCode).saveAs("returnCode")
)
)
).doIf(Session::isFailed).then(
exec(session -> {
log.error("Failed request seq: {} ; Return seq: {} ; Return code: {} ; session: {}",
session.getString("seq"),
session.getString("returnSeq"),
session.getString("returnCode"),
session.userId()
);
return session.reset();
}
)
)
).exec(ws("Close WS").close());
public WebsocketTest() {
setUp(
scn.injectOpen(rampUsers(connections).during(rampUp))
).protocols(httpConf);
}
}
```
But the result is so confusing. In this pic, validation is passed, but an error msg is still logged.
So `Session::isFailed` returns true?
And here looks like thread gatling-1-2 consumes gatling-1-3’s msg, then gatling-1-3 has to wait for nothing until timeout and fails the request.
Env:
- Gradle 7.5 w/ Gatling plugin 3.8.3.2
- JDK 11
- macOS 12.1
---
### Author:  [@slandelle](https://community.gatling.io/u/slandelle)
#### Post date: [August 18, 2022, 1:45pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/4 "2022-08-18T13:45:22Z")
> [@nobody](#):
>
> ```auto
> session -> {
> session.markAsSucceeded();
> String nanoTime = String.valueOf(System.nanoTime());
>
> ```
Session is immutable: [Gatling - Session API](https://gatling.io/docs/gatling/reference/current/core/session/session_api/#setting-attributes).
---
### Author:  [@nobody](https://community.gatling.io/u/nobody)
#### Post date: [August 18, 2022, 1:57pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/5 "2022-08-18T13:57:50Z")
Thank you slandelle. The 1st problem is solved, but the 2nd one still here.
New code:
```java
.exec(session -> {
Session newSession = session.markAsSucceeded();
String nanoTime = String.valueOf(System.nanoTime());
return newSession.set("seq", System.currentTimeMillis() + nanoTime.substring(nanoTime.length() - 5));
})
```
I got 12 out of 3500 requests being mistakenly treated as non-matching:
```auto
================================================================================
2022-08-18 21:51:01 121s elapsed
---- Requests ------------------------------------------------------------------
> Global (OK=8388 KO=12 )
> Connect WS (OK=700 KO=0 )
> Send text (OK=3500 KO=0 )
> Check response code (OK=3488 KO=12 )
> Close WS (OK=700 KO=0 )
---- Errors --------------------------------------------------------------------
> Check Check response code timeout 12 (100.0%)
---- WebSocket -----------------------------------------------------------------
[##########################################################################]100%
waiting: 0 / active: 0 / done: 700
================================================================================
```
---
### Author:  [@slandelle](https://community.gatling.io/u/slandelle)
#### Post date: [August 18, 2022, 2:25pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/6 "2022-08-18T14:25:15Z")
Your `matchJsonPath` (missing from your sample) is likely wrong.
```java
.matching(
jsonPath(matchJsonPath).ofLong().isEL("#{seq}").saveAs("returnSeq")
)
```
Note that saveAs here won’t work (in matching).
---
### Author:  [@nobody](https://community.gatling.io/u/nobody)
#### Post date: [August 18, 2022, 2:35pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/7 "2022-08-18T14:35:30Z")
`matchJsonPath` is using the default value and unchanged during the test.
If the jsonpath is wrong, all requests should fail.
```java
String matchJsonPath = System.getProperty("matchJsonPath", "$.body.seq");
```
The response body is fixed (except client seq(`$.body.seq`), server seq(`$.seq`) and timestamp), so `$.body.seq` is always available.
```json
{"body":{"code":10001,"desc": "UNSUPPORT PROTOCOL","ptype":"UNRECOGNIZED","seq":166083059168627347},"ptype":"ACK","seq":1660830593473,"timestamp":1660830593474, "ver":"1.0.0"}
```
---
### Author:  [@nobody](https://community.gatling.io/u/nobody)
#### Post date: [August 18, 2022, 2:42pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/8 "2022-08-18T14:42:42Z")
Under higher concurrency, more and more responses seemed to fail to match the “seq” attribute in Session?
---
### Author:  [@slandelle](https://community.gatling.io/u/slandelle)
#### Post date: [August 18, 2022, 2:50pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/9 "2022-08-18T14:50:18Z")
> session.set(“seq”, System.currentTimeMillis() + nanoTime.substring(nanoTime.length() - 5));
“seq” is a String here.
> jsonPath(matchJsonPath).ofLong().isEL(“#{seq}”).saveAs(“returnSeq”)
And here you’re comparing a long and a String, so it’s bound to fail.
---
### Author:  [@slandelle](https://community.gatling.io/u/slandelle)
#### Post date: [August 18, 2022, 2:51pm UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/10 "2022-08-18T14:51:01Z")
Then, if you face an issue, please provide a reproducer as instructed here: [How to Ask a Question](https://community.gatling.io/t/how-to-ask-a-question/6538).
---
### Author:  [@nobody](https://community.gatling.io/u/nobody)
#### Post date: [August 24, 2022, 6:25am UTC](https://community.gatling.io/t/need-help-strange-behavior-of-websocket-checks/7080/11 "2022-08-24T06:25:23Z")
> And here you’re comparing a long and a String, so it’s bound to fail.
I found out the root cause, nothing to do with test code or gatling.
There was a bug in our server code, in some circumstances, server sends msg to a wrong channel.