Tair (Redis® OSS-Compatible):Retry mechanisms for Redis clients

Retry only idempotent operations

A timeout can occur at any stage of command execution:

• The command has not reached the server.

• The command reached the server, but execution timed out.

• The command executed on the server, but the response timed out.

Because a retried operation may execute more than once, only retry idempotent operations.

Idempotent -- SET a b: Running this command multiple times always produces the same result.

Non-idempotent -- LPUSH mylist a: Running this command multiple times appends duplicate a elements to mylist.

Configure appropriate retry count and interval

Set the retry count and interval based on your workload:

• Too few retries or too long an interval: The application may fail to complete operations during brief transient faults.

• Too many retries or too short an interval: The application consumes excessive resources and may overwhelm the server.

Avoid retry nesting

Retry nesting -- where a retry operation triggers another retry loop -- can cause repeated or unlimited retries. Implement retry logic at a single layer in your application stack.

Log retry failures

Generate retry logs at the WARN level. Log only when the final retry fails, not on each individual attempt, to avoid log noise.

Standard or cluster instance in proxy mode (JedisPool)

For standard instances or cluster instances in proxy mode, use PooledConnectionProvider with UnifiedJedis.

This example retries the SET command up to 5 times within 10 seconds, with exponentially increasing wait times. If all retries fail, an exception is thrown.

PooledConnectionProvider provider = new PooledConnectionProvider(HostAndPort.from("127.0.0.1:6379"));
int maxAttempts = 5; // Maximum number of retries
Duration maxTotalRetriesDuration = Duration.ofSeconds(10); // Maximum total retry duration
UnifiedJedis jedis = new UnifiedJedis(provider, maxAttempts, maxTotalRetriesDuration);
try {
    System.out.println("set key: " + jedis.set("key", "value"));
} catch (Exception e) {
    // The operation failed after maxAttempts retries or after maxTotalRetriesDuration elapsed.
    e.printStackTrace();
}

Cluster instance in direct connection mode (JedisCluster)

For cluster instances in direct connection mode, use JedisCluster. The maxAttempts parameter sets the number of retry attempts. The default value is 5. If the operation fails after the maximum number of attempts, an exception is thrown.

HostAndPort hostAndPort = HostAndPort.from("127.0.0.1:30001");
int connectionTimeout = 5000;
int soTimeout = 2000;
int maxAttempts = 5;
ConnectionPoolConfig config = new ConnectionPoolConfig();
JedisCluster jedisCluster = new JedisCluster(hostAndPort, connectionTimeout, soTimeout, maxAttempts, config);
try {
    System.out.println("set key: " + jedisCluster.set("key", "value"));
} catch (Exception e) {
    // The operation failed after maxAttempts retries.
    e.printStackTrace();
}