[Tarantool-patches] [RFC] gc/xlog: delay xlog cleanup until relays are subscribed

Vladislav Shpilevoy v.shpilevoy at tarantool.org
Fri Mar 19 02:04:04 MSK 2021



On 18.03.2021 22:31, Cyrill Gorcunov wrote:
> On Thu, Mar 18, 2021 at 09:54:12PM +0100, Vladislav Shpilevoy wrote:
>>>>
>>>> Then it should have 'replication_' prefix, not 'wal_'. Because
>>>> it is ignored if replicas connect before the timeout expires.
>>>
>>> Replication is one of the reason while main gamer is "wal"
>>> here. In the series I sent recently I named it "wal_cleanup_delay".
>>> In future we might introduce some topology detector as you've
>>> been suggesting and better to not stick to "replication"
>>> prefix I think.
>>
>> The thing you said is just another argument for it having 'replication'
>> prefix. Because topology also is not about WAL.
> 
> Topology is anoter _reason_ to defer wal cleanup, and there might
> be other reasons unrelated to replications at all.

Topology is a part of replication. It **is** replication.

If there would be other reasons, your option won't work, because
replication simply drops it when all nodes are connected.

> On the other
> hands I see what you mean and won't insist. Should I name it
> replication_cleanup_delay then?

I don't know yet. I would call it replication_max_gc_delay probably.
But you need to ask Mons for a final decision. Perhaps he will
insist on wal_cleanup_delay. See my other points in the v2 thread.


More information about the Tarantool-patches mailing list