- **apiserver_flowcontrol_priority_level_seat_count_samples**: histograms of seats occupied by executing requests (both regular and final-delay phases included), broken down by priority_level; the observations are taken once per millisecond.

   */
  @GwtIncompatible // Duration
  @SuppressWarnings("GoodTime") // duration decomposition
  private static long toNanosSaturated(Duration duration) {
    // Using a try/catch seems lazy, but the catch block will rarely get invoked (except for
    // durations longer than approximately +/- 292 years).
    try {
      return duration.toNanos();
    } catch (ArithmeticException tooBig) {

                fileSize = response.getDataSize();

                // this is so damn unreliable, needs another race-prone query if required
                haveAttributes = false;

                // This seems to be the only way to obtain a reliable (with respect to locking) file size here
                // It is more critical than other attributes because append mode depends on it.

flip(data, options) {\n  // if `inner` modifier is enabled, we can't use the `flip` modifier\n  if (isModifierEnabled(data.instance.modifiers, 'inner')) {\n    return data;\n  }\n\n  if (data.flipped && data.placement === data.originalPlacement) {\n    // seems like flip is trying to loop, probably there's not enough space on any of the flippable sides\n    return data;\n  }\n\n  const boundaries = getBoundaries(\n    data.instance.popper,\n    data.instance.reference,\n    options.padding,\n    options....

0F,YAAYlB,GACtB6B,EAASN,IAAI,CACXC,OAAQ,UACRC,MAAO,YAELI,EAASnE,SAASqC,IACpB8B,EAASX,YAAYnB,GAGvB8B,EAASC,aAGX5E,KAAKC,SAASnB,QAAQR,EAAAA,QAAEoD,MA7HP,4BA6H+B1B,KAAKE,YAGvD6E,eAAA,WACM/E,KAAKE,QAAQM,SAASsC,GACxB9C,KAAK6E,WAIP7E,KAAKoE,cAKPxC,MAAA,SAAMoD,GAAM,IAAAC,EAAAjF,KACVA,KAAKE,QAAU8E,EAEf1G,EAAAA,QAAE0B,MAAMe,KAAKf,KAAKK,UAAU8C,iBAAiB+B,OAAM,WACjDD,EAAKd,YAGP7F,EAAAA,QAAE0B,MAAMe,KAAKf,KAAKK,UAAUgD,iBAAiB6B,OAAM,WACjDD,EAAKF,oBAGPzG,EAAAA,QAAE0B,MAAMe,KAAKf,KAAKK,UAAU+C,eAAe8B,OAAM,WAC/CD,...

    if (expectedSize < 3) {
      checkNonnegative(expectedSize, "expectedSize");
      return expectedSize + 1;
    }
    if (expectedSize < Ints.MAX_POWER_OF_TWO) {
      // This seems to be consistent across JDKs. The capacity argument to HashMap and LinkedHashMap
      // ends up being used to compute a "threshold" size, beyond which the internal table

    if (expectedSize < 3) {
      checkNonnegative(expectedSize, "expectedSize");
      return expectedSize + 1;
    }
    if (expectedSize < Ints.MAX_POWER_OF_TWO) {
      // This seems to be consistent across JDKs. The capacity argument to HashMap and LinkedHashMap
      // ends up being used to compute a "threshold" size, beyond which the internal table

      iptables, saving a bit of memory/time/cpu. (#106373, @aojea) [SIG Network]
- Watch requests that are delegated to aggregated apiservers no longer reserve concurrency units (seats) in the API Priority and Fairness dispatcher for their entire duration. (#105827, @benluddy) [SIG API Machinery]

## Dependencies

### Added
_Nothing has changed._

### Changed

- Updated kube-apiserver's priority & fairness work estimator such that 'max seats' is MIN(0.15 x nominalCL, nominalCL / handSize)
  

- Introduced new apiserver metric `apiserver_flowcontrol_current_inqueue_seats`. This metric is analogous to `apiserver_flowcontrol_current_inqueue_requests`, but tracks the total number of seats, as each request can take more than one seat. ([#119385](https://github.com/kubernetes/kubernetes/pull/119385), [@andrewsykim](https://github.com/andrewsykim))