Yukun Cao

base case一个建议 “base case 为 dp[0][..] = 0， dp[..][0] = 1。因为如果不使用任何硬币面值，就无法凑出任何金额；如果凑出的目标金额为 0，那么“无为而治”就是唯一的一种凑法。” 这段不是很好理解，尤其是后半句dp[..][0] = 1 无为而治太抽象了 我的做法是dp[0][0]=1 其余dp[0][..] = 0 不选任何硬币当然可以凑出面值为0的方案，且只有一种，而不能凑成面值>0的方案。 而剩下的dp[..][0]交给迭代去处理不是更好吗？ ``` int n = coins.length; int[][] dp =...

@kbaichoo Hi, suggested by @htuch from conversation on slack channel, it would be great if you could take a look on the issue, thanks!

> And could you check your `shrink_count` to ensure the shrink action is triggered. Thanks. Thanks for the hint! But where can I use the `shrink_count` parameter to check? Is...

> max Heap Size is: 30MB heapShrinkPercent: 0.5 heapStopAcceptConnPercent: 0.7 > > Your application heap size ranges from 10MB - 22MB from the data provided, and for most of the...

> There are structures that get created from traffic e.g. stats that did exist in the initial state; as such we'd never go back to the initial heap size. I...

Thanks for the explaination! Still I am afraid that my questions still exists: 1. Overall it's not some small O(10)MBs use cases, I was testing with O(10Mb) because it's easy...

> In my view "used==current_allocated_bytes + fragmentation - unmapped" is more reasonable > > > Yes I agree, we should subtract `Stats::totalPageHeapFree` in the fixed heap monitor's computation of resource...

@KBaichoo any further idea or suggestions? thanks

Hi @KBaichoo thanks for the PR, I am not sure/afraid it won't solve my issue though. Ofc I will do more tests to verify. 1. my question above was asking...

Hi @KBaichoo my test results meet the expectation that the issue still persists. 1. "More broadly, to avoid your deployment from freezing up, you could consider some additional overload manager...