There are multiple ways, historically, to interpret a d20 check in D&D.
The first is "this is how well I tried this". In this interpretation, DCs are objective facts about how hard something is. Anything with no consequences to failure that is easier than 20+Modifiers on the player's part can be defeated by repeated attempts.
The second is "this is how hard the task really is". A DC of 25 isn't an objective fact about the difficulty of opening this door, but rather a description of how hard doors vaguely like this are to open.
When you roll your d20 plus your strength modifier, you are seeing "are you strong enough to break down this door in particular". If you fail, the answer is "no, you aren't strong enough". And if you aren't strong enough, attempting again doesn't make you stronger.
This second interpretation is the "you get one try to pick a lock, and if you fail, you cannot try again until you gain a level" one. Similarly, someone can try to break a door again if they find a way to make themselves significantly better at knocking down doors (get a better tool, a strength buff, whatever).
A third way has come to D&D from other games. The idea is that every check should have stakes and consequences, either explicitly told to the player, or in the mind of the DM.
Before you call for a check, determine the DC. Then determine what happens if the check fails or succeeds.
If the answer to a failure is "nothing happens, and the player can try again", then don't ask for a roll. The rolls do nothing of interest.
Instead, you can do this; first, you determine that it is possible. Then you set a DC. If the player beats the DC, the door is bashed open. If it fails, they are told "it is taking a while and it is noisy, but you can do it ... eventually. Do you want to continue until it breaks?" If they say yes, you might (throw a wandering encounter at them) or (do a check for a wandering encounter) or (increase alertness in some foes nearby); this should be planned before you ask for the roll, at least vaguely.
Then don't ask for another roll; instead, they break down the door and the fiction continues.
In some game systems you tell the player the stakes ahead of time: "DC 25, if you fail you will take longer and make a lot of noise and hostile creatures may respond. If you succeed, it is noisy but fast." In others, you just track the consequences yourself.
This can also be known as "fail forward"; a failure should advance the plot.
Now, failure doesn't have to be successful. An option of failure is "bashing the door causes the archway to collapse on top of you, take XdY damage and the entrance is now buried under rubble". The point is failure should cause something to happen to change the state of the game.
Under 1, ask them to roll once. If they succeed, it opens immediately. If not, check 20+modifier vs DC; if it wins, they break it down eventually. Don't bother asking for more rolls; maybe roll to determine how long it takes.
Under 2, they roll once. If they fail, they may not try again until they boost their chances.
Under 3, you don't let them roll until you know what happens when they fail. Then they roll, and either the success or failure result happens.
Under all 3 of these cases, having someing roll repeatedly doesn't make sense, especially out of combat. In combat, seeing how many rounds it takes can be extremely interesting, so you might want to still use repeated rolls.