If one day lag sums up to 100G only, your disks really suck... wow! That would be a bit more than 1MB/s.
Well, you have your solution I guess 😏
the keyword replica identity doesn't even supported in hypertables exactly, that's why I wanna know if there are any workaround
Читать полностью…If the problem is that you need a replica identity, but REPLICA IDENTITY FULL isn't working, you can set the REPLICA IDENTITY to any unique index using the ALTER TABLE ... REPLICA IDENTITY ... command. The documentation I linked to explains how to do this.
Читать полностью…hypertables in timescaledb doesn't support that. it is just supported within native postgreSQL table
Читать полностью…How to do that? I have just the id, and time column as my primary key and that's it
Читать полностью…any solution pls? I have only the insert working fine, but update and delete needing replicate identity.... any other solution or work around???
Читать полностью…By "db to not work" I assume you have a lot of blocking transactions?
If you don't need to update the records atomically, you can do something like
WITH to_update AS (
SELECT ctid FROM your_table
WHERE your_condition
LIMIT 10000
)
UPDATE your_table
SET whatever
FROM to_update
WHERE your_table.ctid = to_update.ctid;
UPDATE 0
back.
Читать полностью…
In a few of the articles I have gone through .. Pg 9.6 to 12 after that 12 to 16 is good practice because significant changes happen between 9.6 to 16 .
Читать полностью…I think "postgres: startup recovering 000000010000BDC70000003F 99.99 io" in iostat reason to grow up pg_wal, but i dont understand why so much io..
Читать полностью…That's not telling anything. Check archive_mode
and if that's set to always
check pg_stat_archiver
...
on cpu and ram both equal, but disk, i think is garbage... lag grow up to one day ^^'
Читать полностью…If that doesn't work, you should probably direct the question to a Timescale resource rather than a PostgreSQL community resource, as you will find more people there with experience in Timescale.
Читать полностью…that's exactly what I'm trying to explain to you that alter table tablename replica identity full, this doesn't work with hypertable
Читать полностью…Probably nothing to do but create a synthetic primary key (probably just a bigint serial).
Читать полностью…Hey guys, I have a timescale hypertable, all I want is to have a trigger on insert, update, delete operations to replicate the rows to other same table structure in the database.
Only the insert statement working fine, but the update and delete operations needing replica identity on the table, but replicate identity full not supported in hypertable as it is only suppprted in native postgres table
Hi everyone
I have a table where I have 2 cr records where I am doing update operation which is leading to db to not work.
Can anyone suggest any efficient way to do update operation on a table with 2 cr data
There is only one way for minimum downtime and it's logical replica + switchover. All other methods are with increasing amount of downtime. First decide how much downtime is tolerable then ask for "minimum".
Читать полностью…Here i have one question .. is pg_upgrade work on 9.6 to 12 ? or what is the best way to upgrade with minimum downtime. Your thoughts will be appreciated 🙏
Читать полностью…The replay happens in a single process, while your primary probably uses many more. Yeah, that would be an obvious reason.
Secondary significantly less powerful than the primary as well, I guess? Maybe even running on spinning rust?
Postgres entrerprice doesn't exist if you are using some postgres fork I recommend you to create a case to support site of your "postgres version".
One posible reason to pg_wal grows up is archive mode has always value and archive process is failing at standby server...there are more possibles issues