Antidote crashing when multi-DC overloaded in a multi-DC setting.

[aletomsic]

Full trace in the file. It looks like it is caused by attempting to read from an ets table that does not exist under the following config.

{recover_meta_data_on_start, true} {enable_logging, false}

The error does not occur when setting both parameters to true or false.

The error occurs in the master branch and can be reproduced in a config with multiple DCs, running basho_bench executing transactions that update a large amount of objects (500 in my case)

multidcCrash.log

[bieniusa]

Does it only happen with a large number of data?

[aletomsic]

My guess is that, at least, the interDC must be saturated to see this happening. The problem does not occur under moderated load (fewer updates). So I guess not.