StackExchange.Redis

Where are KEYS, SCAN, FLUSHDB etc?

Some very common recurring questions are:

There doesn’t seem to be a Keys(...) or Scan(...) method? How can I query which keys exist in the database?

or

There doesn’t seem to be a Flush(...) method? How can I remove all the keys in the database?

The key word here, oddly enough, is the last one: database. Because StackExchange.Redis aims to target scenarios such as cluster, it is important to know which commands target the database (the logical database that could be distributed over multiple nodes), and which commands target the server. The following commands all target a single server:

• KEYS / SCAN
• FLUSHDB / FLUSHALL
• RANDOMKEY
• CLIENT
• CLUSTER
• CONFIG / INFO / TIME
• REPLICAOF
• SAVE / BGSAVE / LASTSAVE
• SCRIPT (not to be confused with EVAL / EVALSHA)
• SHUTDOWN
• SLOWLOG
• PUBSUB (not to be confused with PUBLISH / SUBSCRIBE / etc)
• some DEBUG operations

(I’ve probably missed at least one) Most of these will seem pretty obvious, but the first 3 rows are not so obvious:

• KEYS / SCAN only list keys that are on the current server; not the wider logical database
• FLUSHDB / FLUSHALL only remove keys that are on the current server; not the wider logical database
• RANDOMKEY only selects a key that is on the current server; not the wider logical database

Actually, StackExchange.Redis spoofs the RANDOMKEY one on the IDatabase API by simply selecting a target server at random, but this is not possible for the others.

So how do I use them?

Simple: start from a server, not a database.

// get the target server
var server = conn.GetServer(someServer);

// show all keys in database 0 that include "foo" in their name
foreach(var key in server.Keys(pattern: "*foo*")) {
    Console.WriteLine(key);
}

// completely wipe ALL keys from database 0
server.FlushDatabase();

Note that unlike the IDatabase API (where the target database has already been selected in the GetDatabase() call), these methods take an optional parameter for the database, or it defaults to 0.

The Keys(...) method deserves special mention: it is unusual in that it does not have an *Async counterpart. The reason for this is that behind the scenes, the system will determine the most appropriate method to use (KEYS vs SCAN, based on the server version), and if possible will use the SCAN approach to hand you back an IEnumerable<RedisKey> that does all the paging internally - so you never need to see the implementation details of the cursor operations. If SCAN is not available, it will use KEYS, which can cause blockages at the server. Either way, both SCAN and KEYS will need to sweep the entire keyspace, so should be avoided on production servers - or at least, targeted at replicas.

So I need to remember which server I connected to? That sucks!

No, not quite. You can use conn.GetEndPoints() to list the endpoints (either all known, or the ones specified in the original configuration - these are not necessarily the same thing), and iterate with GetServer() to find the server you want (for example, selecting a replica).

This site is open source. Improve this page.