This document explains how to manually trigger a major compaction in a Spanner database.
Several storage-related features in Spanner, such as tier storage or columnar engine, rely on a database-wide major compaction before they are fully enabled. By default, major compactions occur automatically across all tables over seven day periods. This means you might wait up to seven days for a new feature to be fully available. To make new features available immediately, you can manually trigger a major compaction.
The compaction process is a long-running operation (LRO) .
Pricing
Manually triggering a major compaction temporarily increases CPU utilization on your Spanner instance. If additional compute capacity is required to offset this increase in CPU utilization, this will increase costs.
Performance
Major compactions run as background operations. However, if your instance has consistently heavy CPU usage, the compaction workload might interfere with other critical operations. In such cases, you can temporarily scale up the instance to ensure stable performance during compaction.
Manually trigger a major compaction
Google Cloud console
-
Open the Google Cloud console and select your instance.
-
Select a database.
-
In the navigation menu, click Spanner Studio.
-
Open a new tab by clicking New SQL editor tabor New tab.
-
Run one of the following commands to start compaction, based on your database dialect:
GoogleSQL
CALL compact_all ();PostgreSQL
CALL spanner . compact_all ();This operation returns a long-running operation (LRO) ID that you can use to find the operation in the Operationslist.
-
To monitor the progress of the compaction operation, in the navigation menu, click Operations.
C++
To trigger compactions programmatically using the C++ client library:
void
Compact
(
google
::
cloud
::
spanner
::
Client
client
)
{
namespace
spanner
=
::
google
::
cloud
::
spanner
;
spanner
::
SqlStatement
select
(
"CALL compact_all()"
);
using
RowType
=
std
::
tuple<std
::
string
> ;
auto
rows
=
client
.
ExecuteQuery
(
std
::
move
(
select
));
for
(
auto
&
row
:
spanner
::
StreamOf<RowType>
(
rows
))
{
if
(
!
row
)
throw
std
::
move
(
row
).
status
();
std
::
cout
<<
"Long-running operation ID: "
<<
std
::
get<0>
(
*
row
)
<<
"
\n
"
;
}
}
You can check the progress of a long-running database operation . You can also cancel the ongoing major compaction request using the LRO ID. For more information, see Cancel a long-running database operation .
