mnesia: intensive usage table

Question

When I receive such message, like this:

** WARNING ** Mnesia is overloaded: {dump_log, write_threshold}

what is approach to known which table is being used intensively? What kind of debugging do I need to do?

Thanks.

Bianca

Hard to say what you should do. Do you have many writes to the same row? which processes use mnesia and how much? Why do you hit the dump-log threshold? – Etiology

Added to my todo list to learn the cyrillic alphabet. – Aleta

How can I know what request (operation with mnesia) has locked table if it happend? – Bianca

The {dump_log, write_threshold} happens when the dump_log gets written too often. Perhaps you can turn the knob dump_log_write_threshold up. It defaults to 100. – Etiology

Answer 1

Interesting post over on StreamHacker on the subject:

http://streamhacker.com/2008/12/10/how-to-eliminate-mnesia-overload-events/

One of the suggestions in the article is switching to synchronous writes, not sure that this is such a good generic answer. It would certainly change your scaling characteristics, so don't skip engineering and measuring the change.

However both dc_dump_limit and dump_log_write_threshold are good settings to play with. The defaults are somewhat overly sensitive.

Blackheart