WordPress can store the results of database queries in memory so that repeated requests do not hit the database again. That is object caching, and on a busy site it makes a real difference.
What is available
Our servers provide memcached. It is available to your account and works well as an object cache.
What is not
Redis is not available on our shared servers. This matters because of how it fails. The Redis client library is present, so a plugin will happily install and appear to work – it gets as far as trying to connect, and only then fails. WordPress falls back to its internal cache, so the site keeps working and nothing looks wrong on screen.
What you get instead is an exception written to your error log on every single request. We have seen an account accumulate thousands of these into a multi-megabyte log file without the owner noticing anything.
The catch that makes this hard to spot
A Redis cache leaves behind a file called object-cache.php in wp-content. WordPress loads that file whether or not any plugin is still installed. So you can uninstall the Redis plugin, see a clean plugins screen, and still be generating an error on every request, because the drop-in file was left behind.
What to do
If you have arrived from another host and are seeing error log growth, check for wp-content/object-cache.php. If it is a Redis drop-in, remove it. Then either use memcached or rely on page caching, which for most sites gives the bigger improvement anyway.
