Re: [Tarantool-patches] [PATCH v8 1/2] applier: send transaction's first row WAL time in the applier_writer_f

[Serge Petrenko via Tarantool-patches <tarantool-patches@dev.tarantool.org>]

16.06.2021 16:32, Cyrill Gorcunov пишет:
> On Tue, Jun 15, 2021 at 12:36:02PM +0300, Serge Petrenko wrote:
>>> @@ -193,6 +196,16 @@ applier_writer_f(va_list ap)
>>>    			applier->has_acks_to_send = false;
>>>    			struct xrow_header xrow;
>>>    			xrow_encode_vclock(&xrow, &replicaset.vclock);
>>> +			/*
>>> +			 * For relay lag statistics we report last
>>> +			 * written transaction timestamp in tm field.
>>> +			 *
>>> +			 * Replica might be dead already so we have to
>>> +			 * test on each iteration.
>>> +			 */
>>> +			struct replica *r = replica_by_id(replica_id);
>>> +			if (likely(r != NULL))
>>> +				xrow.tm = r->applier_txn_start_tm;
>> How could a replica be dead here?
>> AFAIR we delete a replica only when it's deleted from _cluster. Shouldn't
>> the applier writer be dead as well by that time?
> Before accessing replica_by_id we're sitting in event loop trying to
> fetch data from the network. Which means an admin may cleanup the
> entry manually before we get back to this code in result we will
> get a nil dereference (if only I'm not missing something).

I mean that this code you're referring to is executed in a
while (!fiber_is_cancelled()) {...} loop.

IIRC applier->writer gets cancelled as soon as replica is unregistered.
So even if someone deletes the entry manually the writer will exit the loop
before getting a nil dereference.

P.S. I couldn't find this anywhere in code so let's leave the check.

-- 
Serge Petrenko