wp cache flush breaks when using redis cluster on 2.5.3

[Trisky]

Description

wp cache flush breaks when using redis cluster on 2.5.3 and predis

Expected Behavior

cache cleared

Actual Behavior

Flush fails

wp cache flush
Predis\NotSupportedException: The command 'INFO' is not allowed in replication mode. in /var/html/www/vendor/predis/predis/src/Replication/ReplicationStrategy.php:51
Stack trace:
#0 /var/html/www/vendor/predis/predis/src/Connection/Replication/SentinelReplication.php(547): Predis\Replication\ReplicationStrategy->isReadOperation(Object(Predis\Command\Redis\INFO))
#1 /var/html/www/vendor/predis/predis/src/Connection/Replication/SentinelReplication.php(581): Predis\Connection\Replication\SentinelReplication->getConnectionInternal(Object(Predis\Command\Redis\INFO))
#2 /var/html/www/vendor/predis/predis/src/Connection/Replication/SentinelReplication.php(713): Predis\Connection\Replication\SentinelReplication->getConnectionByCommand(Object(Predis\Command\Redis\INFO))
#3 /var/html/www/vendor/predis/predis/src/Connection/Replication/SentinelReplication.php(753): Predis\Connection\Replication\SentinelReplication->retryCommandOnFailure(Object(Predis\Command\Redis\INFO), 'executeCommand')
#4 /var/html/www/vendor/predis/predis/src/Client.php(381): Predis\Connection\Replication\SentinelReplication->executeCommand(Object(Predis\Command\Redis\INFO))
#5 /var/html/www/vendor/predis/predis/src/Client.php(336): Predis\Client->executeCommand(Object(Predis\Command\Redis\INFO))
#6 /var/html/www/web/app/object-cache.php(1153): Predis\Client->__call('info', Array)
#7 /var/html/www/web/app/object-cache.php(562): WP_Object_Cache->fetch_info()
#8 /var/html/www/web/app/object-cache.php(256): WP_Object_Cache->__construct(true)
#9 /var/html/www/web/wp/wp-includes/load.php(860): wp_cache_init()
#10 /var/html/www/web/wp/wp-settings.php(144): wp_start_object_cache()
#11 phar:///usr/local/bin/wp/vendor/wp-cli/wp-cli/php/WP_CLI/Runner.php(1374): require('/var/html/www/w...')
#12 phar:///usr/local/bin/wp/vendor/wp-cli/wp-cli/php/WP_CLI/Runner.php(1293): WP_CLI\Runner->load_wordpress()
#13 phar:///usr/local/bin/wp/vendor/wp-cli/wp-cli/php/WP_CLI/Bootstrap/LaunchRunner.php(28): WP_CLI\Runner->start()
#14 phar:///usr/local/bin/wp/vendor/wp-cli/wp-cli/php/bootstrap.php(83): WP_CLI\Bootstrap\LaunchRunner->process(Object(WP_CLI\Bootstrap\BootstrapState))
#15 phar:///usr/local/bin/wp/vendor/wp-cli/wp-cli/php/wp-cli.php(32): WP_CLI\bootstrap()
#16 phar:///usr/local/bin/wp/php/boot-phar.php(20): include('phar:///usr/loc...')
#17 /usr/local/bin/wp(4): include('phar:///usr/loc...')
#18 {main}
Error: The object cache could not be flushed.

Possible Fix

Steps to Reproduce

1. Enable cluster with WP_REDIS_DISABLE_GROUP_FLUSH=true
2. run wp cache flush
3. error

Additional context

Problem is fixed as soon as I rollback to v2.5.2

Environment

• Plugin version: 2.5.3
• PHP version: 8.2
• WordPress version: 6.5.5

[tillkruss]

Can you try the develop branch which includes #541?

[erfan-ilyas]

Got the same issue.

Got these constants defined.

define( 'WP_REDIS_IGBINARY', true );
define( 'WP_REDIS_DISABLE_METRICS', true );
define( 'WP_REDIS_DISABLE_BANNERS', true );
define( 'WP_REDIS_DISABLE_ADMINBAR', true );
define( 'WP_REDIS_DISABLE_DROPIN_CHECK', true );
define( 'WP_REDIS_DISABLE_DROPIN_AUTOUPDATE', true );

define( 'WP_REDIS_CLIENT', 'predis' );
define( 'WP_REDIS_SENTINEL', 'mymaster' );
define( 'WP_REDIS_SERVERS', ['tcp://127.0.0.1:26379',] );

Was working fine on Version 2.5.0

I will try the version 2.5.2 as stated by @Trisky

[tillkruss]

@erfan-ilyas I just merged #544, does that resolve it for you?

[dustinrue]

@tillkruss I am seeing that #544 seems to cause this:

Predis\ClientException: Command `GETMASTER` is not a registered Redis command. in /var/www/html/wordpress/wp-content/plugins/redis-cache/dependencies/predis/predis/src/Command/Factory.php:67
Stack trace:
#0 /var/www/html/wordpress/wp-content/plugins/redis-cache/dependencies/predis/predis/src/Client.php(313): Predis\Command\Factory->create()
#1 /var/www/html/wordpress/wp-content/plugins/redis-cache/dependencies/predis/predis/src/Client.php(304): Predis\Client->createCommand()
#2 /var/www/html/wordpress/wp-content/object-cache.php(1151): Predis\Client->__call()
#3 /var/www/html/wordpress/wp-content/object-cache.php(562): WP_Object_Cache->fetch_info()

If I revert this change it appears to work fine and I do not get OPs original issue about the info command. I am using this configuration:

define( 'WP_REDIS_CLIENT', 'predis' );
define( 'WP_REDIS_SENTINEL', 'mymaster' );
define( 'WP_REDIS_SERVERS', ['tcp://redis:26379/?password=ZfY2jy4W7xwA3spcJhHRtQ'] );
define( 'WP_REDIS_PASSWORD', 'ZfY2jy4W7xwA3spcJhHRtQ' );
define( 'WP_REDIS_TIMEOUT', 1 );
define( 'WP_REDIS_READ_TIMEOUT', 1 );
define( 'WP_REDIS_DATABASE', 0 ); 

I am running Redis Sentinal in Kube using the bitnami/redis chart. I only need to spec a single sentinel because the Kube load balancer will always route me to a live sentinel which is replicated. I am running the develop branch as is with that commit backed out.

[tillkruss]

@yatsukhnenko Can you take another look at this?

[neverlless]

+1