Quarkus Redis client timeout shorter than a second

[abutic]

Describe the bug
The docs says Redis client timeout is defined as a Duration:
https://quarkus.io/guides/redis#quarkus-redis-client_quarkus.redis.timeout

A Duration object is later passed to an underlying implementation, as can be seen here:

quarkus/extensions/redis-client/runtime/src/main/java/io/quarkus/redis/client/runtime/RedisClientImpl.java

Line 1026 in 6dce423

io.vertx.mutiny.redis.client.Response response = mutinyResponse.await().atMost(Duration.ofSeconds(timeout));

but, internally, only seconds precision is used:

quarkus/extensions/redis-client/runtime/src/main/java/io/quarkus/redis/client/runtime/RedisClientImpl.java

Line 15 in 6dce423

public RedisClientImpl(RedisAPI redisAPI, long timeout) {

, with the timeout parameter representing the seconds.

So, without fractional seconds precision, we lost the ability to have timeouts shorter than a second.

Expected behavior
Instead of silently discarding fractional seconds part, which later leads to a misleading error when client is used:

java.lang.IllegalArgumentException: `duration` must be greater than zero

, it would be nice to allow timeouts shorter than a second, by storing the Duration internally and later passing it as is to Mutiny Redis Client.

The alternative is to throw a more precise exception, saying for instance that timeouts shorter than a second, even though they're allowed by the Duration, are not allowed.

To Reproduce
Define a timeout shorter than a second and try using a RedisClient. E.g.:

quarkus.redis.timeout=PT0.1S

[quarkus-bot]

/cc @cescoffier, @gsmet, @machi1990

[gsmet]

@abutic would you be interested in providing a pull request? Looks like you did most of the work already :).

[abutic]

@abutic would you be interested in providing a pull request? Looks like you did most of the work already :).

I was "afraid" you'd ask... :)

PR: #15211

[abutic]

@gsmet and @machi1990, should I close this issue, now that it's fixed by #15211? Also, should the PR be linked to this issue (I don't seem to have the permission to link them)?

[geoand]

@abutic I'll close it.

Just as an FYI for next time, if you add Fixes: #NNNNN like I have done in the description of your PR, GH will automatically link the PR to the issue :)