PostgreSQL 13 finally introduced the possibility to change the `primary_conninfo` without a restart. Just doing reload is enough, but in case if the role is changing from the `replica` to the `standby_leader` we want to call only `on_role_change` callback and skip `on_reload`, because they duplicate each other.
We don't need to rewind when:
1. replayed location for the former replica is not ahead of switchpoint
2. end of checkpoint record for the former primary is the same as switchpoint
In order to get the end of checkpoint record we use the `pg_waldump` and parse its output.
Close https://github.com/zalando/patroni/issues/1493
The standby cluster doesn't know about leader elections in the main cluster and therefore the usual mechanisms of detecting divergences don't work. For example, it could happen that the standby cluster is ahead of the new primary of the main cluster and must be rewound.
There is a way to know that the new timeline has been created by checking the presence of a history file in pg_wal. If the new file is there, we will start usual procedures of making sure that we can continue streaming or will run the pg_rewind.
The `touch_member()` could be called from the finally block of the `_run_cycle()`. In case if it raised an exception the whole Patroni process was crashing.
In order to avoid future crashes we wrap `_run_cycle()` into the try..except block and ask a user to report a BUG.
Close https://github.com/zalando/patroni/issues/1529
So far Patroni was parsing `recovery.conf` or querying `pg_settings` in order to get the current values of recovery parameters. On PostgreSQL earlier than 12 it could easily happen that the value of `primary_conninfo` in the `recovery.conf` has nothing to do with reality. Luckily for us, on PostgreSQL 9.6+ there is a `pg_stat_wal_receiver` view, which contains current values of `primary_conninfo` and `primary_slot_name`. The password field is masked through, but this is fine, because authentication happens only during opening the connection. All other parameters we compare as usual.
Another advantage of `pg_stat_wal_recevier` - it contains the current timeline, therefore on 9.6+ we don't need to use the replication connection trick if walreceiver process is alive.
If there is no walreceiver process available or it is not streaming we will stick to old methods.
It is safe to call pg_rewind on the replica only when pg_control on the primary contains information about the latest timeline. Postgres is usually doing immediate checkpoint right after promote and in most cases it works just fine. Unfortunately we regularly receive complaints that it takes to long (minutes) until the checkpoint is done and replicas can't perform rewind. At the same time doing the checkpoint manually immediately helped. So Patroni starts doing the same. When the promotion happened and postgres is not running in recovery, we explicitly issue the checkpoint.
We are intentionally not using the AsyncExecutor here, because we want the HA loop continues doing its normal flow.
## Feature: Postgres stop timeout
Switchover/Failover operation hangs on signal_stop (or checkpoint) call when postmaster doesn't respond or hangs for some reason(Issue described in [1371](https://github.com/zalando/patroni/issues/1371)). This is leading to service loss for an extended period of time until the hung postmaster starts responding or it is killed by some other actor.
### master_stop_timeout
The number of seconds Patroni is allowed to wait when stopping Postgres and effective only when synchronous_mode is enabled. When set to > 0 and the synchronous_mode is enabled, Patroni sends SIGKILL to the postmaster if the stop operation is running for more than the value set by master_stop_timeout. Set the value according to your durability/availability tradeoff. If the parameter is not set or set <= 0, master_stop_timeout does not apply.
That required a refactoring of `Config` and `Patroni` classes. Now one has to explicitely create the instance of `Config` before creating `Patroni`.
The Config file can optionally call the validate function.
We will try to import only the module which has a configuration section.
I.e. if there is only zookeeper section in the config, Patroni will try to import only `patroni.dcs.zookeeper` and skip `etcd`, `consul`, and `kubernetes`.
This approach has two benefits:
1. When there are no dependencies installed Patroni was showing INFO messages `Failed to import smth`, which looks scary.
2. It reduces memory usage, because sometimes dependencies are heavy.
Starting from PostgreSQL 12 the following recovery parameters could be changed without restart, but Patroni didn't yet support it:
* archive_cleanup_command
* promote_trigger_file
* recovery_end_command
* recovery_min_apply_delay
In future postgres releases this list will be extended and Patroni will support it automatically.
Specifically, there was a chance that `patronictl reinit --force` was overwritten by recover and we end up in a situation when Patroni was trying to start the postgres while basebackup still running.
* make it possible to use client certificates with REST API
* define a separate PatroniRequest class which handles all communication
* refactor patronictl to use the new class
* make Ha to use the new class instead of calling requests.get. The old call wasn't taking into account certificates and basic-auth
Close#898
* Convert postgresql.py into a package
* Factor out cancellable process into a separate class
* Factor out connection handler into a separate class
* Move postmaster into postgresql package
* Factor out pg_rewind into a separate class
* Factor out bootstrap into a separate class
* Factor out slots handler into a separate class
* Factor out postgresql config handler into a separate class
* Move callback_executor into postgresql package
This is just a careful refactoring, without code changes.
1. use the default port is 5432 when only standby_cluster.host is defined
2. check that standby_cluster replica can be bootstrapped without connection to the standby_cluster leader against `create_replica_methods` defined in the `standby_cluster` config instead of the `postgresql` section.
3. Don't fallback to the create_replica_methods defined in the `postgresql` section when bootstrapping a member of the standby cluster.
4. Make sure we specify the database when connecting to the leader.
* expose the current patroni version in DCS
* expose `checkpoint_after_promote` flag in DCS as an indicator that pg_rewind could be safely executed
* other nodes will wait until this flag is set instead of connecting as superuser and issuing the CHECKPOINT
* define `postgresql.authention.rewind` with credentials for pg_rewind in patroni configuration files.
* create user for pg_rewind if postgres is 11+
* grant execute on functions required for pg_rewind to rewind user
This functionality works similarly to the `pg_hba`:
If the `postgresql.pg_ident` is defined in the config file or DCS, Patroni will write its value to pg_ident.conf, however, if `postgresql.parameters.ident_file` is defined, Patroni will assume that pg_ident is managed from outside and not update the file.
First of all, this patch changes the behavior of `on_start`/`on_restart` callbacks, they will be called only when postgres is started or restarted without role changes. In case if the member is promoted or demoted only the `on_role_change` callback will be executed. `on_role_change` was never called for standby leader, only `on_start`/`on_restart` and with a wrong role argument.
Before that `on_role_change` was never called for standby leader, only `on_start`/`on_restart` and with a wrong role argument.
In addition to that, the REST API will return standby_leader role for the leader of the standby cluster.
Closes https://github.com/zalando/patroni/issues/988
It might happen that the standby cluster is configured to be created and replay WAL files from the different source than when it is running not in standby mode. This is necessary to avoid writing WAL files and backups into the old place after promotion.
The easiest way to achieve such behavior is passing RemoteMember object to `Postgresql.clone` method instead of the usual Member object.
If the pg_rewind is disabled or can't be used, the former master could fail to start as a new replica due to diverged timelines. In this case, the only way to fix it is wiping the data directory and reinitializing.
So far Patroni was able to remove the data directory only after failed attempt to run pg_rewind. This commit fixes it.
If the `postgresql.remove_data_directory_on_diverged_timelines` is set, Patroni will wipe the data directory and reinitialize the former master automatically.
Fixes: https://github.com/zalando/patroni/issues/941
The latest timeline is calculated from the `/history` key in DCS. In case there is no such key or it contains some garbage we consider the node healthy.
Closes https://github.com/zalando/patroni/issues/890
Permanent replication slots are preserved on failover/switchover, that is Patroni on the new primary will create configured replication slots right after doing promote.
Slots could be configured with the help of `patronictl edit-config`.
The initial configuration could be also done in the `bootstrap.dcs`
```yaml
slots:
permanent_physical_1:
type: physical
permanent_logical_1:
type: logical
database: foo
plugin: pgoutput
```
It is the responsibility of the operator to make sure that there are no clashes in names between replication slots automatically created by Patroni for members and permanent replication slots.
Closes https://github.com/zalando/patroni/issues/656
* Always run `pg_rewind` against the remote master
* Always use the remote master as the source when "recovering" stopped standby leader
* Use remote master as the source when "recovering" the node in the unhealthy cluster
* Use the local dbname as the fallback when doing `pg_rewind` from the remote master
* `no_replication_slot` is the allowed key in the `RemoteMember` object
* Make it possible to "bootstrap" the new `standby_cluster` with existing (and valid) data directory. There is one prerequisite though, there should be no `patroni.dynamic.json` file in it!
* Use `shutil.move` instead of `os.replace`, which is available only from 3.3
* Introduce standby-leader health-check and consul service
* Improve unit tests, some lines were not covered
* rename `assertEquals` -> `assertEqual`, due to deprecation warning
Implementation of "standby cluster" described in #657. Standby cluster consists
of a "standby leader", that replicates from a "remote master" (which is not a
part of current patroni cluster and can be anywhere), and cascade replicas,
that replicate from the corresponding standby leader. "Standby leader" behaves
pretty much like a regular leader, which means that it holds a leader lock in
DSC, in case if disappears there will be an election of a new "standby
leader".
One can define such a cluster using the section "standby_cluster" in patroni
config file. This section provides parameters for standby cluster, that will be
applied only once during bootstrap and can be changed only through DSC.
Do not exit when the cluster system ID is empty or the one that doesn't pass the validation check. In that case, the cluster most likely needs a reinit; mention it in the result message.
Avoid terminating Patroni, as otherwise reinit cannot happen.
Patroni can attach itself to an already running PostgreSQL instance. If that is the first instance "seen" in the given cluster, Patroni for that instance will create the initialize key, grab the leader key and, if the instance is running a replica, promote.
Because of this behavior, when a cluster with a master and one or more replicas gets Patroni for each node, it is imperative to start running Patroni on the master node before getting to the replicas.
This commit changes such weird behavior and will abort Patroni start if there is no initialize key in DCS and postgres is running as a replica.
Closes https://github.com/zalando/patroni/issues/655
It is very easy to get current timeline on the master by executing
```sql
SELECT ('x' || SUBSTR(pg_walfile_name(pg_current_wal_lsn()), 1, 8))::bit(32)::int
```
Unfortunately the same method doesn't work when postgres is_in_recovery. Therefore we will use replication connection for that on the replicas. In order to avoid opening and closing replication connection on every HA loop we will cache the result if its value matches with the timeline of the master.
Also this PR introduces a new key in DCS: `/history`. It will contain a json serialized object with timeline history in a format similar to the usual history files. The differences are:
* Second column is the absolute wal position in bytes, instead of LSN
* Optionally there might be a fourth column - timestamp, (mtime of history file)
Make it possible to cancel a running task if you want to reinitialize replica.
There are two possible ways to trigger it:
1. patronictl will ask whether you want to cancel already running task if an attempt to trigger reinitialize has failed
2. if you are using `--force` argument with `patronictl reinit`
Introduces a PostmasterProcess object that identifies a running process via pid and start time.
When pid file is parsed and the correct process identified this object is passed around.
When the process goes away we try to find a new one in case somebody restarted postgres behind our back.
It could happen that ttl provided in Patroni configuration is smaller
than minimum supported by Consul. In such case Consul agent fails to
create a new session and responds with 500 Internal Server Error and
http body contains something like: "Invalid Session TTL '3000000000',
must be between [10s=24h0m0s]". Without session Patroni is not able to
create member and leader keys in the Consul KV store and it means that
cluster becomes completely unhealthy.
As a workaround we will handle such exception, adjust ttl to the minimum
possible and retry session creation.
In addition to that make it possible to define custom log format via environment variable `PATRONI_LOGFORMAT`
When Patroni does calculation whether it should run pg_rewind or not, it relies on pg_controldata output or gets necessary information from replication connection.
On some cases (when for example postgres running as a master was killed), we can't use pg_controldata output immediately, but trying to start postgres. Such start could fail with the following errror:
```
LOG,00000,"ending log output to stderr",,"Future log output will go to log destination ""csvlog"".",,,,,,,""
LOG,00000,"database system was interrupted; last known up at 2017-09-16 22:35:22 UTC",,,,,,,,,""
LOG,00000,"restored log file ""00000006.history"" from archive",,,,,,,,,""
LOG,00000,"entering standby mode",,,,,,,,,"" 2017-09-18 08:00:39.433 UTC,,,57,,59bf7d26.39,4,,2017-09-18 08:00:38 UTC,,0,LOG,00000,"restored log file ""00000006.history"" from archive",,,,,,,,,""
FATAL,XX000,"requested timeline 6 is not a child of this server's history","Latest checkpoint is at 29/1A000178 on timeline 5, but in the history of the requested timeline, the server forked off from that timeline at 29/1A000140.",,,,,,,,""
LOG,00000,"startup process (PID 57) exited with exit code 1",,,,,,,,,""
LOG,00000,"aborting startup due to startup process failure",,,,,,,,,""
LOG,00000,"database system is shut down",,,,,,,,,""
```
In this case controldata will still have `Database cluster state: in production`
All further attempts to start postgres will fail. Such situation could be fixed only if we start not in recovery. For safety we will do it in a single user mode.
The second problems is: if postgres was running as master, but later we started it and stopped, than pg_controldata will report:
```
Database cluster state: shut down in recovery
Minimum recovery ending location: 0/0
Min recovery ending loc's timeline: 0
```
And this info can't be used for calculations. In this case we should use
`Latest checkpoint location` and `Latest checkpoint's TimeLineID`
in addition to that implement additional checks around manual failover and recover when synchronous_mode is enabled
* Comparison must be case insensitive
