[DOC] Add details on the conjunction of innodb_corrupt_table_action and innodb_force_recovery

[Jean Weisbuch]

In case of corruption of an InnoDB table (using files-per-table), there are two variables that are here to try to recover/salvage/remove it :
    - innodb_force_recovery
    - innodb_corrupt_table_action

The first is the most common one (from upstream InnoDB) while the second one is an XtraDB specific, both variables settings might be overlapping if a corrupted InnoDB table is encountered, the documentation is not clear on which takes precedence.

What should happen if innodb_force_recovery is set to >0 (allows the the server to keep running even if corrupt pages are detected) and you try to access a corrupted InnoDB table while innodb_corrupt_table_action is set to :
    - assert (intentionally crash the server when it detects corrupted data in a single-table tablespace) (the default value) : should it crash the server with an assertion or let it try to read from table?
    - warn (disable all further I/O (except for deletion) on the table file) : should it be possible to try to read from these table or only to DROP them?

ps: does "salvage" only applies if "innodb_file_per_table" is enabled, as for "warn" or in any case?

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PS-1543