# Être derrière un proxy { #behind-a-proxy }

Dans de nombreuses situations, vous utiliserez un **proxy** comme Traefik ou Nginx devant votre application FastAPI.

Ces proxies peuvent gérer les certificats HTTPS et d'autres aspects.

## En-têtes transférés par le proxy { #proxy-forwarded-headers }

	}

	// Verify if we can employ autotune without running out of capacity,
	// if we do run out of capacity, make sure to turn-off autotuning
	// in such situations.
	if autotune {
		newConcurrent := concurrent + (concurrent+1)/2
		autoTunedCapacityNeeded := uint64(newConcurrent * size)
		if capacity < autoTunedCapacityNeeded {

* vSphere Volume Plugin Implementation ([#24947](https://github.com/kubernetes/kubernetes/pull/24947), [@abithap](https://github.com/abithap))
* ResourceQuota controller uses rate limiter to prevent hot-loops in error situations ([#25748](https://github.com/kubernetes/kubernetes/pull/25748), [@derekwaynecarr](https://github.com/derekwaynecarr))

Il y a normalement un autre outil chargé de **démarrer et exécuter** votre conteneur.

Cela pourrait être **Docker** directement, **Docker Compose**, **Kubernetes**, un **service cloud**, etc.

Dans la plupart (ou toutes) des situations, il existe une option simple pour activer l'exécution du conteneur au démarrage et activer les redémarrages en cas d'échec. Par exemple, dans Docker, c'est l'option de ligne de commande `--restart`.

Au final, dans les deux situations, il est fort probable que **FastAPI** soit tout de même [plus rapide](index.md#performance) que (ou au moins de vitesse égale à) votre framework précédent.

### Dépendances { #dependencies }

* Bug fixes

  * Changed Kubelet default image-gc-high-threshold to 85% to resolve a conflict with default settings in docker that prevented image garbage collection from resolving low disk space situations when using devicemapper storage. ([#40432](https://github.com/kubernetes/kubernetes/pull/40432), [@sjenning](https://github.com/sjenning))

   ([#122530](https://github.com/kubernetes/kubernetes/pull/122530), [@neolit123](https://github.com/neolit123))...

parking-citations-10.json.zst {"Ticket":1103341116,"IssueData":"2015-12-21T00:00:00","IssueTime":"1251","RPState":"CA","PlateExpiry":"200304","Make":"HOND","BodyStyle":"PA","Color":"GY","Location":"13147 WELBY WAY","Route":"01521","Agency":1,"ViolationCode":"4000A1","ViolationDescr":"NO EVIDENCE OF REG","Fine":50,"Latitude":99999,"Longitude":99999} {"Ticket":1103700150,"IssueData":"2015-12-21T00:00:00","IssueTime":"1435","RPState":"CA","PlateExpiry":"201512","Make":"GMC","BodyStyle":"VN","Col...

first update to position the popper in the right place\n    this.update();\n\n    const eventsEnabled = this.options.eventsEnabled;\n    if (eventsEnabled) {\n      // setup event listeners, they will take care of update the position in specific situations\n      this.enableEventListeners();\n    }\n\n    this.state.eventsEnabled = eventsEnabled;\n  }\n\n  // We can't use class properties because they don't get listed in the\n  // class prototype and break stuff like Sinon stubs\n  update() {\n  ...

The precise details are relevant for Go implementations,
affect the specifics of error messages (such as whether
a compiler reports a type inference or other error),
and may explain why type inference fails in unusual code situations.
But by and large these rules can be ignored when writing Go code:
type inference is designed to mostly "work as expected",
and the unification rules are fine-tuned accordingly.
</p>

<p>