[avargas]

By a quick glance, they are only logging 10% of exceptions, so if there's a lot, logs don't get clogged.

[jstanley]

Also, the first exception is logged regardless.

[bsg75]

So, log first exception, then a random 10% sample of the following instances of the same exception?

IIF the approach of limiting the number of logged exceptions is valid, does it make any sense to randomly choose which ones - other than it was an easy one-line hack?