Re: Avoid full page images in streaming replication?
От | Magnus Hagander |
---|---|
Тема | Re: Avoid full page images in streaming replication? |
Дата | |
Msg-id | CABUevExEt3y+vOBXzKHNsrfbibKpXmA6d=xj=eyn4PHx7QkBMQ@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: Avoid full page images in streaming replication? (Tom Lane <tgl@sss.pgh.pa.us>) |
Ответы |
Re: Avoid full page images in streaming replication?
|
Список | pgsql-hackers |
<p dir="ltr"><br /> On Oct 22, 2015 23:38, "Tom Lane" <<a href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>> wrote:<br/> ><br /> > Jim Nasby <Jim.Nasby@BlueTreble.com> writes:<br /> > > ISTM it should be possibleto avoid sending full page writes to a<br /> > > streaming replica once the replica has reached a consistentstate. I<br /> > > assume that the replica would still need to write full pages to it's<br /> > > diskin case of a crash, but the sender could insert special WAL records<br /> > > to tell it when to do so, insteadof sending the full page image.<br /> > > Presumably this would be a big win for replication over a WAN.<br/> ><br /> > How are you going to make that work without LSNs in the WAL received by<br /> > the replicadiverging from those in the master's WAL?<br /> ><p dir="ltr">We could in theory send a "this would be been a fpibut it's skipped" record which would only exist in streaming and just make the standby write a noop of some kind? It wouldstill be on the standby but it would at least not consume the bandwidth. Just skip sending the actual contents of thefpi. <p dir="ltr">/Magnus
В списке pgsql-hackers по дате отправления: