Group flush not working on AWS Elasticache Serverless

Are you seeing an error when you’re running wp_cache_flush_group()?

@julieadrienne it just returns false. Here are our cache settings in wp-config.php:

define('WP_REDIS_SCHEME',          'tls');
define('WP_REDIS_HOST',            'AWS_ENDPOINT_REDACTED');
define('WP_REDIS_PORT',            6379);
define('WP_REDIS_DISABLE_BANNERS', true);
define('WP_REDIS_IGNORED_GROUPS', ['groups']); // this is the group that we're looking to flush separately

Let me know if I need to enable any debugging option to get more verbose output.

Are those all the WP_REDIS_* constants you’re setting, or are there any others?

Does your AWS instance have cluster mode enabled or disabled?

That’s all the constants we’re setting, everything else is as is.

Elasticache Serverless is cluster mode enabled by default I believe. There is no option in the admin panel. From the announcement the client has to support cluster mode but they don’t provide a configuration endpoint.

Can you dump redis-cli INFO CLUSTER in here to see what sort of cluster AWS Elasticache Serverless is using?

Here you go, with a couple extra commands. Let me know if you need anything else.

INFO:

# Server
redis_version:7.1
redis_mode:cluster
arch_bits:64
run_id:0

# Replication
role:master
connected_slaves:1
slave0:ip=REDACTED,port=6380,state=online,offset=0,lag=0

# Cluster
cluster_enabled:1

INFO CLUSTER:

# Cluster
cluster_enabled:1

CLUSTER INFO:

cluster_state:ok
cluster_slots_assigned:16384
cluster_slots_ok:16384
cluster_slots_pfail:0
cluster_slots_fail:0
cluster_known_nodes:2
cluster_size:1
cluster_current_epoch:0
cluster_my_epoch:0
cluster_stats_messages_sent:0
cluster_stats_messages_received:0
total_cluster_links_buffer_limit_exceeded:0

Great. You’re using a native Redis Cluster, that means you cannot use the WP_REDIS_HOST configuration, you must use:

define( 'WP_REDIS_CLIENT', 'phpredis' );

define( 'WP_REDIS_CLUSTER', [
    'tcp://127.0.0.1:6379?alias=node-01',
    'tcp://127.0.0.2:6379?alias=node-02',
    'tcp://127.0.0.3:6379?alias=node-03',
] );

Hi @julieadrienne,

I did give that a try but I’m getting this error: Couldn't map cluster keyspace using any provided seed

CLUSTER SHARDS returns the following:

1) 1) slots
   2) 1) (integer) 0
      2) (integer) 16383
   3) nodes
   4) 1)  1) id
          2) 391fe7f8456c55dd9da927432439baa600000000
          3) ip
          4) 127.0.0.1
          5) endpoint
          6) REDACTED
          7) hostname
          8) REDACTED
          9) role
         10) master
         11) replication-offset
         12) 0
         13) health
         14) online
         15) tls-port
         16) 6379
      2)  1) id
          2) a3745ba00b485da78396197c0cf99a5400000000
          3) ip
          4) 127.0.0.1
          5) endpoint
          6) REDACTED
          7) hostname
          8) REDACTED
          9) role
         10) replica
         11) replication-offset
         12) 0
         13) health
         14) online
         15) tls-port
         16) 6380

I tried both endpoints in both TLS and TCP mode and get the same error every time.

Appreciate your help!

What does your config look like now?

I tried the following configs, all preceded with the phpredis line, unsuccessfully:

define( 'WP_REDIS_CLUSTER', [
    'tcp://FIRST_ENDPOINT:6379?alias=node-01',
    'tcp://SECOND_ENDPOINT:6380?alias=node-02',
] );

define( 'WP_REDIS_CLUSTER', [
    'tls://FIRST_ENDPOINT:6379?alias=node-01',
    'tls://SECOND_ENDPOINT:6380?alias=node-02',
] );

define( 'WP_REDIS_CLUSTER', [
    'tcp://FIRST_ENDPOINT:6379?alias=node-01',
] );

define( 'WP_REDIS_CLUSTER', [
    'tls://FIRST_ENDPOINT:6379?alias=node-01',
] );

define( 'WP_REDIS_CLUSTER', [
    'tcp://SECOND_ENDPOINT:6380?alias=node-01',
] );

define( 'WP_REDIS_CLUSTER', [
    'tls://SECOND_ENDPOINT:6380?alias=node-01',
] );

I’ll note that we’ve been running the non-cluster config (I added it a couple comments above) with no issues apart from the group flush specifically.

• This reply was modified 10 months, 3 weeks ago by alexl75.

Please post the full configuration you’re using and the error message you’re seeing with that configuration.

With this configuration, everything works except group flush:

define('WP_REDIS_CLIENT', 'phpredis');
define('WP_REDIS_SCHEME', 'tls');
define('WP_REDIS_HOST', 'AWS_ENDPOINT_REDACTED');
define('WP_REDIS_PORT', 6379);
define('WP_REDIS_DISABLE_BANNERS', true);
define('WP_REDIS_IGNORED_GROUPS', ['groups']); // this is the group that we're looking to flush separately

With this configuration, I get the following error on every page load: Couldn't map cluster keyspace using any provided seed

define('WP_REDIS_CLIENT', 'phpredis');
define('WP_REDIS_CLUSTER', [
    'tls://AWS_ENDPOINT_REDACTED:6379?alias=node-01',
]);
define('WP_REDIS_DISABLE_BANNERS', true);

Try removing the WP_REDIS_SCHEME and WP_REDIS_HOST and WP_REDIS_PORT.

You only need the WP_REDIS_CLUSTER.

I did try the 2 configurations I indicated as their own, separately with no other variables.

With WP_REDIS_HOST everything works fine except group flush.

With WP_REDIS_CLUSTER I get this error Couldn't map cluster keyspace using any provided seed on every page load.

If your Redis cluster shows cluster_enabled:1, then you MUST use WP_REDIS_CLUSTER.

Using WP_REDIS_HOST with a native Redis Cluster doesn’t work.