2830
English speaking PostgreSQL public chat. This group is for discussions on PostgreSQL-related topics and strives to provide best-effort support as well.
Yeah, except that the data will come from the secondary
Читать полностью…
btw - im aware of logical replication
im just playing around with physical replication now
How can I achieve something like this? if it's even possible?
I thought in case it's part of replication slot it will save the last state to replicate from
Removing the standby.signal will make the timelines diverge
Читать полностью…
Thanks, Jamie, it looks like 1 database per postgresql cluster is the best option.
Читать полностью…
Hello everyone,
I’m new to PostgreSQL and would like to ask a beginner question.
Is it considered best practice to use a single PostgreSQL cluster to host multiple databases for different applications? From what I understand, all databases in a cluster share the same WAL stream, so I’m unclear how point-in-time recovery could be performed for an individual database within that cluster.
Also, RDBMS can be as fast as their transaction logs, are multiple databases add performance overhead to the single transaction log file compared just to have a single db on a particular cluster?
Did OP delete the question? Or am I just a bit dumb right now? 😅
Читать полностью…
Feel free to use it, it might work, it would probably work
Читать полностью…
The question is extremely meaningless.
Don't ask questions, write code.
(No, you cannot make architectural blueprints in areas you don't know anything about. Just write the system.)
For example, consider something like Signal: it's very secure, they keep barely any data (and such data can also be sharded very effectively)
Читать полностью…
so this would be like just creating the table on the current master , no?
Читать полностью…
You can, however, promote the secondary, create your tables, then make the primary a standby and sync it, then make the primary again primary, and switch back once again the secondary to make it secondary again
Читать полностью…
There is no way to merge back to primary after the standby diverges, unless you use something like pg_rewind but that moves you back in time, it doesn't integrate the changes
Читать полностью…
Is it possible to remove the standby.signal from replica in order to create some tables only for replica, and then touch standby.signal and start the replica again? will it work? or it will damage the WAL order?
Читать полностью…
Well in that case you will waste a little of ram because shared_buffer is assigned by one cluster and it tends to be a large memory segment. Every case has its drawbacks and your best choice will depend on your use case
Читать полностью…
Point in time recovery impacts to the whole instance so you would recover every database.
Читать полностью…
I wouldn't use it because i don't like SaaS because of the added latency + complexity (postgresql is fairly easy and self contained to use), but some peoople like the trade off of paying money to not have to care about that
Читать полностью…
A good relational database intro book would give you some general info.
Читать полностью…
The best way to keep data safe is to have as least of it as possible, or as Germans would say, "datensparsamkeit"
Читать полностью…
PgBouncer is great if you need it. Which problem are you trying to fix?
Читать полностью…