我试图在一个相对复杂的Node.js服务器应用程序中分析内存/ GC问题.即使在非常适中的负载下,它也会在明显的时期内变得无法响应,并且随着时间的推移这些停顿会变得更长.使用--trace-gc参数运行表明可能导致极长的垃圾回收时间:
[4805] 537 ms: Mark-sweep 17.6 (46.4) -> 10.3 (47.4) MB, 20 ms [allocation failure] [GC in old space requested].
[4805] 1338 ms: Mark-sweep 31.3 (58.4) -> 19.2 (57.2) MB, 40 ms [allocation failure] [promotion limit reached].
[4805] 2662 ms: Mark-sweep 58.0 (79.2) -> 43.9 (85.2) MB, 109 ms [Runtime::PerformGC] [promotion limit reached].
[4805] 4014 ms: Mark-sweep 90.1 (111.5) -> 70.6 (113.9) MB, 114 ms [allocation failure] [promotion limit reached].
[4805] 7283 ms: Mark-sweep 129.7 (153.9) -> …