disregard. problem found. i was locking the wrong table.
On 9/16/2019 11:10 AM, Kevin Wilkinson wrote:
on linux, pg11.3, i have a (declarative) partitioned table with a
deadlock that i do not understand. one process does a copy to the
partitioned table. another process is executing a jdbc
on linux, pg11.3, i have a (declarative) partitioned table with a
deadlock that i do not understand. one process does a copy to the
partitioned table. another process is executing a jdbc batch of commands
to "atomically" replace one of the table partitions. it has the
following commands (autoco
we were able to use a brin index for our time-series data and that saved
a lot of space. basically, we used a btree for recent data and, once the
data was "stable", reclustered the data to get high correlation on the
brin index, created a brin index and dropped the btree. it works well.
On 4/1
for updating a single row.
kevin
On 4/7/2019 11:41 PM, Laurenz Albe wrote:
Kevin Wilkinson wrote:
on 10.2, we're seeing very high cpu usage when doing an update statement
on a relatively small table (1GB). one of the updated columns is text,
about 1k bytes. there are four threads doing si
on 10.2, we're seeing very high cpu usage when doing an update statement
on a relatively small table (1GB). one of the updated columns is text,
about 1k bytes. there are four threads doing similar updates
concurrently to the same table (but different rows). each thread does an
update about ever
i want to recluster an immutable table without locking the table and
then add the table as a partition of a parent table. my plan was:
create table tbl_cpy as select * from tbl order by c1 asc;
alter table tbl_cpy add constraint c1 not null, c1>=lo c1i want this to run quickly. so, i popul
if you are able/willing to use ZFS (rather than ext4, xfs, ...) to store
your database, then it might work for you. ZFS is copy-on-write so it
can very quickly clone a database.
kevin
On 3/1/2019 12:08 PM, Arjun Ranade wrote:
I'm working on a project that requires on-demand creation of a fresh
i have a range partitioned table with a brin index that i am using for
"Internet-of-Things" kind of data (essentially timeseries data about
some entities). the partition key is a timestamp. data is only loaded to
the "current" partition and data is never modified. older partitions are
static. t
