Comment 3 for bug 385717

Thanks for addressing this.

> There are a couple of things I don't like in the changeset you've shown.
> First there seems to be an unrelated change in cbStartAStore (I think I remember where it comes from).

No, it was related. The purpose was for replicating code that called back requesters after successful reconnect in _eb_put().

> Then, I'm not sure why errors in rollback should be treated differently than any other errors. IIRC, the store should reconnect nicely, unless I'm missing something.

You're right. Looking over the code, this shouldn't be a problem, so my _eb_put() was redundant.

So this patch looks good to me. Thanks.