在Jenkins CI崩溃中运行VSCode扩展的“npm测试”

我有一个Jenkins CI设置来运行我的VS代码扩展的测试。我按照VS代码文档中的持续集成页面中的详细信息设置容器https://code.visualstudio.com/api/working-with-extensions/continuous-integration.

apt update
apt install -y libasound2 libgbm1 libgtk-3-0 libnss3 xvfb

然后，我按如下方式运行测试

xvfb-run -a npm run test

直到最近，我的测试开始比以前长了一点。现在，我在测试运行的一半时间内收到以下错误消息。

16:27:34 [main 2023-02-06T15:27:34.720Z] CodeWindow: renderer process gone (reason: crashed, code: 133)
16:27:34 
16:27:34 [main 2023-02-06T15:27:34.723Z] [UtilityProcess id: 1, type: extensionHost, pid: 413]: waiting to exit...
16:27:34 
16:27:34 [main 2023-02-06T15:27:34.736Z] [UtilityProcess id: 1, type: extensionHost, pid: 413]: received exit event with code 0
16:27:34 [main 2023-02-06T15:27:34.737Z] Extension host with pid 413 exited with code: 0, signal: unknown.
16:27:34 
16:27:34 Error sending from webFrameMain:  Error: Render frame was disposed before WebFrameMain could be accessed
16:27:34     at s.send (node:electron/js2c/browser_init:173:631)
16:27:34     at g.send (node:electron/js2c/browser_init:169:2657)
16:27:34     at E.send (/tmp/code/.vscode-test/vscode-linux-x64-1.75.0/VSCode-linux-x64/resources/app/out/vs/code/electron-main/main.js:16:77328)
...

并且测试运行失败。这似乎与测试无关，因为它可能在测试运行中的稍微不同的时间发生。如果运行VS代码测试花费的时间太长，那么似乎会失败。

我该怎么做才能避免这种情况？