What could cause a table to 'crash' and should be repaired? (3 replies)

Hi,

I have a table with about 1Million rows and almost a Gb in size.
I have plenty of space on the server but only 2Gb of memory, (for Apache, php, MySQL etc).

I am using version 5.5.27

This is the table structure.

CREATE TABLE `bhost_cache` (
`cache_key` varchar(255) NOT NULL,
`cache_time` int(10) unsigned NOT NULL,
`cache_data` longblob NOT NULL,
UNIQUE KEY `unique_key` (`cache_key`),
KEY `idx_key_update` (`cache_key`,`cache_time`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

I do some updates from time to time and a lot of selects.

The largest blob is 1Mb but the vast majoriy of the cache is ~500K

Almost every day the table is marked as 'crashed' and I have to run a repair on it.
As it is a cache table I can tell by the sudden load on the server as no cache is been using and the data is recreated dynamically.

My host, the forums, everybody is telling me that the issue is probably with the queries and that I need to 'optimize my queries' but the fact remains that apart from a select, (using the Key), and a couple of "Insert-on-duplicate-update", (again using the key), I don't really do much.

So assuming for a minute that the queries are _not_ the issue.

Could the memory be the issue, (2Gb is a bit low I know)?

What else could be the problem? How else could I monitor the table to get to the root of the problem?

Many thanks in advance

Simon