Release notes for repmgr 2.0: repmgr - Replication Manager for PostgreSQL clusters

repmgr 2.0 released

What is repmgr?

repmgr is a set of open source tools that helps DBAs and System administrators manage a cluster of PostgreSQL databases.

By taking advantage of the Hot Standby capability introduced in PostgreSQL 9, repmgr greatly simplifies the process of setting up and managing database with high availability and scalability requirements.

repmgr simplifies administration and daily management, enhances productivity and reduces the overall costs of a PostgreSQL cluster by:

monitoring the replication process;
allowing DBAs to issue high availability operations such as switch-overs and fail-overs

repmgr is production quality software and used widely across the world with PostgreSQL. Many users rely on repmgr to maintain their replication setups and as such we take new releases seriously and mark the level of maturity as a guide for users. As repmgr 2.0 moves towards production we continue to regard the autofailover feature as currently in beta; we expect that to fully mature in the 2.1 release.

2ndQuadrant provides contract support for PostgreSQL that includes both repmgr and the autofailover feature.

Features

Improved Documentation
General refactoring, code quality improvements and stabilization work
Support for daemonizing (-d/--daemonize)
PID file handling (-p/--pid-file)
New config option: monitor_interval_secs
New config option: retry_promote_interval
New config option: logfile
New config option: pg_bindir
New config option: pgctl_options
Add timestamps to log line in stderr
Add a ssh_options parameter
Make CLONE command try to make an exact copy including $PGDATA location
Add detection of master failure
Add the notion of a witness server
Add experimental autofailover capabilities
Add a configuration parameter to indicate the script to execute on failover or follow
Make the monitoring optional and turned off by default, it can be turned on with --monitoring-history switch
Add tunables to specify number of retries to reconnect to master and the time between them

Bugfixes

Fixed PQexec() calls: fixed several calls where we did not check the result status but only the return value of PQexec(); the query may have failed nonetheless
Flush stderr after a log message appears: We had the problem that the log file appeared empty for a long time due to file buffers. Thus we call fflush() after every log message so the log file gets written out to disk quickly
Fixed repmgr repl_status columns: repmgr repl_status view had the column time_lag which was documented to be the time a standby is behind master. In fact it only works like this when viewed on the standby and not on the master: there it only was the time of the last status update. We dropped that column and replaced it by a new column „communication_time_lag“ which is the content of the repl_status column on the master. On the standby we contain the time of the last update in shared mem though refer always to the correct time nonetheless where repl_status is queried. We also added a new column, „replication_time_lag“, which refers to the apply delay.
Set connections to NULL when calling PQfinish() on them.
Performance improvements: the old implementation took round about 8 seconds per monitoring interval because it got caught in a sleep call and had to wait for timeouts. MUCH too long, especially when you look at the default monitor_interval value of 2 seconds – we could never hold that. The new implementation uses PQgetResult() and select() to avoid the sleep and thus the monitoring routine now only uses a fraction of the time before (<1s).
Leak and memory fixes: Fixed some leaks and an overlapping strcpy() call.
Overhauled CloseConnections(): CloseConnections() did not have a NULL check for PQisBusy() call and was a macro. It also didn't set the connections to NULL. Now it is a function and sets the connections to NULL and checks for NULL before calling functions on connection variables.
Ignore pg_log when cloning
Correctly check wal_keep_segments
General code refactoring
Log format fixes
handle stdin/stdout/stderr for repmgrd
Added format checking for printf() like functions
Added forgotten priority value when creating a witness
pg_config is now setable from outsite of the makefile
Split install targets to install_prog and install_ext with doing both as the default
Flush output before calling system()
Initialize variables as sscanf() leaves them untouched upon error
No longer exit when standby connection drops
Several typos have been corrected
Fixed string comparison when reloading config files
Do not create data directory before sanity checks succeeded when creating a witness
Also check if query was successful when registering a new standby
Remove master node earlier so that master register --force succeeds when it is already registered
Do not exit with database in backup mode (pg_start_backup())
Debian control file now accepts PostgreSQL 9.0, 9.1, 9.2 and 9.3
Now compiles with 9.3

Upcoming

Features we are working on in the near future:

timeline increase when a standby gets promoted
A better check which standby did receive most of the data
Respect the fact that a standby can be delayed on purpose a factor in the voting algorithm
include support for delayed standbys

Community and development

repmgr is free and open source software and is licensed under the GPLv3.