[tarantool-patches] Re: [PATCH v5 2/2] replication: force gc to clean xdir on ENOSPC err

Konstantin Osipov kostja at tarantool.org
Tue Jul 10 21:37:07 MSK 2018

* Konstantin Belyavskiy <k.belyavskiy at tarantool.org> [18/07/10 19:19]:
> Rebase to 1.10 - ok.
> Using relay_stop() makes sense only with replica_on_relay_stop(), since
> relay_stop() itself actually do nothing with consumers.
> Regarding replica_on_relay_stop(), replica should be in "orphan" mode
> to avoid assertion in replica_delete(). And also there is a problem with
> monitoring, since replica will leave replication cluster and thus silent the error.
> On other hand, in case of implementation based on removing consumer,
> replica, if being active again, will get an LSN gap and we will see an error.

This not a problem - it will rejoin once rejoin is in the trunk.
> 1. Please give feedback on this section.
> 2. If not using relay_stop(), which branch use as a base 1.9 or 1.10?

> >Could you write a test with two 
> >"abandoned" replicas, each holding an xlog file? 
> Which xlog, the same one or different for each replicas?

Different one.

I know I skipped some questions - let's discuss the rest
separately, hope the above answers help.

Konstantin Osipov, Moscow, Russia, +7 903 626 22 32
http://tarantool.io - www.twitter.com/kostja_osipov

More information about the Tarantool-patches mailing list