mwtian
mwtian
Before, adding a shared object txn to the pending queue, shared object locks must be taken. Otherwise the certificate will not be able to figure out its shared input object...
I will take a look when I have free cycles.
Have the remaining nodes retried their headers?
> > Have the remaining nodes retried their headers? > > Since the "crashed" nodes have acked the headers , then the "healthy" nodes won't have a reason to retry....
> > > > Have the remaining nodes retried their headers? > > > > > > > > > Since the "crashed" nodes have acked the headers , then...
I wonder why there is no `FuturesUnordered::push()` among the capture stack traces. Hopefully with the new execution driver, we can simplify the execution logic i.e. avoiding some complexities in node...
> Overall, this is fantastic! To summarize the important points from my comments: > > * We need to be deliberate about how pending certificates are persisted and when it...
@mystenmark as we discussed yesterday, I looked into how to make sure shared object locks can be written and cleared correctly per transaction, instead of per epoch as I proposed...
Questions: suppose a shared txn S has 4 owned object parents, and no validator currently has processed all 4 of them owned objects, what mechanism will ensure S is executed...
Also IIUC when a validator signs an owned txn the validator must have all parent objs locally.I wonder what happens when the validator does not have the parent objs locally....