lt_standby — supports the creation of a LightDB warm standby server
lt_standby
[option
...] archivelocation
nextwalfile
walfilepath
[restartwalfile
]
lt_standby supports creation of a “warm standby” database server. It is designed to be a production-ready program, as well as a customizable template should you require specific modifications.
lt_standby is designed to be a waiting
restore_command
, which is needed to turn a standard
archive recovery into a warm standby operation. Other
configuration is required as well, all of which is described in the main
server manual (see Section 24.2).
To configure a standby
server to use lt_standby, put this into its
lightdb.conf
configuration file:
restore_command = 'lt_standby archiveDir
%f %p %r'
where archiveDir
is the directory from which WAL segment
files should be restored.
If restartwalfile
is specified, normally by using the
%r
macro, then all WAL files logically preceding this
file will be removed from archivelocation
. This minimizes
the number of files that need to be retained, while preserving
crash-restart capability. Use of this parameter is appropriate if the
archivelocation
is a transient staging area for this
particular standby server, but not when the
archivelocation
is intended as a long-term WAL archive area.
lt_standby assumes that
archivelocation
is a directory readable by the
server-owning user. If restartwalfile
(or -k
)
is specified,
the archivelocation
directory must be writable too.
There are two ways to fail over to a “warm standby” database server when the master server fails:
In smart failover, the server is brought up after applying all WAL
files available in the archive. This results in zero data loss, even if
the standby server has fallen behind, but if there is a lot of
unapplied WAL it can be a long time before the standby server becomes
ready. To trigger a smart failover, create a trigger file containing
the word smart
, or just create it and leave it empty.
In fast failover, the server is brought up immediately. Any WAL files
in the archive that have not yet been applied will be ignored, and
all transactions in those files are lost. To trigger a fast failover,
create a trigger file and write the word fast
into it.
lt_standby can also be configured to execute a fast
failover automatically if no new WAL file appears within a defined
interval.
lt_standby accepts the following command-line arguments:
-c
Use cp
or copy
command to restore WAL files
from archive. This is the only supported behavior so this option is useless.
-d
Print lots of debug logging output on stderr
.
-k
Remove files from archivelocation
so that
no more than this many WAL files before the current one are kept in the
archive. Zero (the default) means not to remove any files from
archivelocation
.
This parameter will be silently ignored if
restartwalfile
is specified, since that
specification method is more accurate in determining the correct
archive cut-off point.
Use of this parameter is deprecated as of
LightDB 8.3; it is safer and more efficient to
specify a restartwalfile
parameter. A too
small setting could result in removal of files that are still needed
for a restart of the standby server, while a too large setting wastes
archive space.
-r
maxretries
Set the maximum number of times to retry the copy command if
it fails (default 3). After each failure, we wait for
sleeptime
* num_retries
so that the wait time increases progressively. So by default,
we will wait 5 secs, 10 secs, then 15 secs before reporting
the failure back to the standby server. This will be
interpreted as end of recovery and the standby will come
up fully as a result.
-s
sleeptime
Set the number of seconds (up to 60, default 5) to sleep between tests to see if the WAL file to be restored is available in the archive yet. The default setting is not necessarily recommended; consult Section 24.2 for discussion.
-t
triggerfile
Specify a trigger file whose presence should cause failover.
It is recommended that you use a structured file name to
avoid confusion as to which server is being triggered
when multiple servers exist on the same system; for example
/tmp/pgsql.trigger.5432
.
-V
--version
Print the lt_standby version and exit.
-w
maxwaittime
Set the maximum number of seconds to wait for the next WAL file, after which a fast failover will be performed. A setting of zero (the default) means wait forever. The default setting is not necessarily recommended; consult Section 24.2 for discussion.
-?
--help
Show help about lt_standby command line arguments, and exit.
lt_standby is designed to work with LightDB 8.2 and later.
LightDB 8.3 provides the %r
macro,
which is designed to let lt_standby know the
last file it needs to keep. With LightDB 8.2, the
-k
option must be used if archive cleanup is
required. This option remains available in 8.3, but its use is deprecated.
LightDB 8.4 provides the
recovery_end_command
option. Without this option
a leftover trigger file can be hazardous.
lt_standby is written in C and has an easy-to-modify source code, with specifically designated sections to modify for your own needs
On Linux or Unix systems, you might use:
archive_command = 'cp %p .../archive/%f' restore_command = 'lt_standby -d -s 2 -t /tmp/pgsql.trigger.5442 .../archive %f %p %r 2>>standby.log' recovery_end_command = 'rm -f /tmp/pgsql.trigger.5442'
where the archive directory is physically located on the standby server,
so that the archive_command
is accessing it across NFS,
but the files are local to the standby (enabling use of ln
).
This will:
produce debugging output in standby.log
sleep for 2 seconds between checks for next WAL file availability
stop waiting only when a trigger file called
/tmp/pgsql.trigger.5442
appears,
and perform failover according to its content
remove the trigger file when recovery ends
remove no-longer-needed files from the archive directory
On Windows, you might use:
archive_command = 'copy %p ...\\archive\\%f' restore_command = 'lt_standby -d -s 5 -t C:\pgsql.trigger.5442 ...\archive %f %p %r 2>>standby.log' recovery_end_command = 'del C:\pgsql.trigger.5442'
Note that backslashes need to be doubled in the
archive_command
, but not in the
restore_command
or recovery_end_command
.
This will:
use the copy
command to restore WAL files from archive
produce debugging output in standby.log
sleep for 5 seconds between checks for next WAL file availability
stop waiting only when a trigger file called
C:\pgsql.trigger.5442
appears,
and perform failover according to its content
remove the trigger file when recovery ends
remove no-longer-needed files from the archive directory
The copy
command on Windows sets the final file size
before the file is completely copied, which would ordinarily confuse
lt_standby. Therefore
lt_standby waits sleeptime
seconds once it sees the proper file size. GNUWin32's cp
sets the file size only after the file copy is complete.
Since the Windows example uses copy
at both ends, either
or both servers might be accessing the archive directory across the
network.
Simon Riggs <simon@2ndquadrant.com>