From PostgreSQL 10, a member of the default roles "pg_monitor" and/or
"pg_read_all_settings" can read pg_settings without requiring superuser
privileges.
Previously, a hint was being emitted about making the repmgr user a
member of one of those groups, but no check for membership was being
made, meaning the check could only be run by a superuser.
This helps provide a better picture of the state of the cluster, i.e.
making it more obvious whether there's been a timeline divergence.
This also provides infrastructure for further improvements in cluster
status display and diagnosis.
Note this is only available in PostgreSQL 9.6 and later as it relies
on the SQL functions for interrogating pg_control, which can be executed
remotely. As PostgreSQL 9.5 will shortly be the only community-supported
version without these functions, it's not worth the effort of trying
to duplicate their functionality.
Previously repmgr would happily clone from whatever server
it found at the provided source server address. We should
ensure that a standby can only be cloned from a node which
is part of the main replication cluster.
This check fetches a list of nodes from the source server,
connects to the first non-witness server it finds, and
compares the system identifiers of the source node and the
node it has connected to. If there is a mismatch, then the
source server is clearly not part of the main replication
cluster, and is most likely the witness server.
This enables us to better determine whether a node is definitively
attached, definitively not attached, or if it was not possible to
determine the attached state.
This functionality enables repmgrd (when running on the primary) to
monitor connected child nodes. It will log connections and disconnections
and generate events.
Additionally, repmgrd can execute a custom script if the number of connected
child nodes falls below a configurable threshold. This script can be used
e.g. to "fence" the primary following a failover situation where a new primary
has been promoted and all standbys are now child nodes of that primary.
In "recovery.conf", the configuration parameter "node_name" is used
as the "application_name" value, which will be truncated by PostgreSQL
to 63 characters (NAMEDATALEN - 1).
repmgr sometimes needs to be able to extract the application name from
pg_stat_replication to determine if a node is connected (e.g. when
executing "repmgr standby register"), so the comparison will fail
if "node_name" exceeds 63 characters.
Previously, repmgrd assumed that during a failover, there would not
already be another primary node. However it's possible a node was
promoted manually. While this is not a desirable situation, it's
conceivable this could happen in the wild, so we should check for
it and react accordingly.
Also sanity-check that the follow target can actually be followed.
Addresses issue raised in GitHub #420.
This is intended to ensure that all nodes have a constant LSN while
making the failover decision.
This feature is experimental and needs to be explicitly enabled with the
configuration file option "standby_disconnect_on_failover".
Note enabling this option will result in a delay in the failover decision
until the WAL receiver is disconnected on all nodes.
This enable selection of the method repmgrd uses to check whether the upstream
node is available. Possible values are:
- "ping" (default): uses PQping() to check server availability
- "connection": executes a query on the connection to check server
availability (similar to repmgr3.x).
With long-running copy operations, it's possible the connection(s) to
the primary/source server may go away for some reason, so recheck
their availability before attempting to reuse.
In a situation where only some standbys are cut off from the primary,
a failover would result in a split brain/split cluster situation,
as it's likely one of the cut-off standbys will promote itself, and
other cut-off standbys (but not all standbys) will follow it.
To prevent this happening, interrogate the other sibiling nodes to
check whether they've seen the primary within a reasonably short interval;
if this is the case, do not take any failover action.
This feature is experimental.
Specifically, if WAL replay is paused *and* WAL is pending replay,
this node cannot be promoted until WAL replay is unpaused. In this
state it is not a suitable promotion candidate in a failover situation.
If WAL replay is paused but WAL is still pending replay, PostgreSQL will ignore
the promote request until WAL replay is unpaused. This may lead to the standby
being promoted at an unpredictable point in time outside of repmgr's
control. Moreover it may not be obvious that this is happening, or why, and
it will appear that an apparently successful promotion attempt has not
actually worked.
To prevent this from happening, repmgr will now refuse to promote the
standy if WAL replay is paused *and* WAL is still pending replay.
GitHub #540.
Eventually we'll want to have this contain the optional replication
info contained in the t_node_info struct, which should then contain a
pointer to a ReplInfo struct.
If the upstream node has changed, for example after "repmgr standby follow"
was manually executed, restart monitoring to ensure repmgrd is monitoring the
correct node.
Previously, if connection URIs were in use and "repmgr standby switchover"
was executed, repmgr would pass the connection URI as-is to the demotion
candidate to execute "repmgr node rejoin". However the presence of
unescaped ampersands in the connection URI was causing the rejoin command
to be incorrectly executed.
Addresses GitHub #525.
repmgr requires that the same "major version" (e.g. 4.3) is present
on all nodes, otherwise - particularly in the case of repmgrd - it's
highly likely things won't work as expected.
Implements part of GitHub #515.
Most of the time we can simply get the version number directly from
the connection handle. Previously it was held in a global variable,
which was an icky way of doing things.
In a few special cases we also need the actual version string, which
is obtained directly from the database.
This is a somewhat convoluted attempt to retrieve the current LSN
of any node, regardless of whether in recovery or not, and if in
recovery, independent of whether streaming or recovering from
archive.
In some circumstances, e.g. while performing a switchover, it is essential
that repmgrd does not take any kind of failover action, as this will put
the cluster into an incorrect state.
Previously it was necessary to stop repmgrd on all nodes (or at least
those nodes which repmgrd would consider as promotion candidates), however
this is a cumbersome and potentially risk-prone operation, particularly if the
replication cluster contains more than a couple of servers.
To prevent this issue from occurring, this patch introduces the ability
to "pause" repmgrd on all nodes wth a single command ("repmgr daemon pause")
which notifies repmgrd not to take any failover action until the node
is "unpaused" ("repmgr daemon unpause").
"repmgr daemon status" provides an overview of each node and whether repmgrd
is running, and if so whether it is paused.
"repmgr standby switchover" has been modified to automatically pause repmgrd
while carrying out the switchover.
See documentation for further details.
Previously, if the server being monitored was not available, repmgrd
would always close the existing connection handle and open a new one.
However, in some cases, e.g. a brief network outage, the existing
connection handle is still good and does not need to be reopened.
This could be particularly problematic if monitoring_history is on,
as this risks leaving orphan sessions on the primary which (given
a sufficiently unstable network) could lead to all available backends
being occupied.
Instead, during an outage we now use a new connection to verify
the server is accessible; if the old connection is still available
(e.g. following a short network interruption) we continue using that;
if not (e.g. the server was restarted), we use the new one.
a switchover.
We've found that this can cause some issues with postgres control
metadata (could be a postgres bug) so best thing is *not* no switchover
if there's a backup taking place.
It's also a bad idea from an architectual point of view, as a switchover
is supposed to be planed, so why perform it when we are taking backups.
GitHub #476.
