* the capacity, it is doubled in size unless some other thread
   * holds the lock. If a hashed slot is empty, and lock is
   * available, a new Cell is created. Otherwise, if the slot
   * exists, a CAS is tried.  Retries proceed by "double hashing",
   * using a secondary hash (Marsaglia XorShift) to try to find a
   * free slot.
   *
   * The table size is capped because, when there are more threads

   * the capacity, it is doubled in size unless some other thread
   * holds the lock. If a hashed slot is empty, and lock is
   * available, a new Cell is created. Otherwise, if the slot
   * exists, a CAS is tried.  Retries proceed by "double hashing",
   * using a secondary hash (Marsaglia XorShift) to try to find a
   * free slot.
   *
   * The table size is capped because, when there are more threads

   * the capacity, it is doubled in size unless some other thread
   * holds the lock. If a hashed slot is empty, and lock is
   * available, a new Cell is created. Otherwise, if the slot
   * exists, a CAS is tried.  Retries proceed by "double hashing",
   * using a secondary hash (Marsaglia XorShift) to try to find a
   * free slot.
   *
   * The table size is capped because, when there are more threads

// operation succeeded.
func (h *healSequence) pushHealResultItem(r madmin.HealResultItem) error {
	// start a timer to keep an upper time limit to find an empty
	// slot to add the given heal result - if no slot is found it
	// means that the server is holding the maximum amount of
	// heal-results in memory and the client has not consumed it
	// for too long.

      int hash = element.hashCode();
      for (int j = Hashing.smear(hash); ; j++) {
        int index = j & mask;
        Object value = table[index];
        if (value == null) {
          // Came to an empty slot. Put the element here.
          elements[uniques++] = element;
          table[index] = element;
          hashCode += hash;
          break;
        } else if (value.equals(element)) {
          break;

    //     fgha[c]|[b]de -> fgha[b]|[c]de. Because these two blocks are the same size, we are done.
    // The Dolphin algorithm is attractive because it does the fewest array reads and writes: each
    // array slot is read and written exactly once. However, it can have very poor memory locality:
    // benchmarking shows it can take 7 times longer than the other two in some cases. The other two

In many cases you might already know the answer for those questions. 🤓

If you are helping a lot of people with their questions, you will become an official [FastAPI Expert](fastapi-people.md#fastapi-experts){.internal-link target=_blank}. 🎉

So, to balance that out, imagine the following short story:

> You have to clean a big, dirty house.

*Yep, that's the whole story*.

---

There's no waiting 🕙 anywhere, just a lot of work to be done, on multiple places of the house.

An interesting detail is that the percentage of the **CPU used** by each process can **vary** a lot over time, but the **memory (RAM)** normally stays more or less **stable**.

If you have an API that does a comparable amount of computations every time and you have a lot of clients, then the **CPU utilization** will probably *also be stable* (instead of constantly going up and down quickly).

So, if your application consumes a lot of memory (for example with machine learning models), and your server has a lot of CPU cores **but little memory**, then your container could end up trying to use more memory than what is available, and degrading performance a lot (or even crashing). 🚨

### Create a `Dockerfile`