the pgpool postgres proxy and load balancer

This chart uses the PGPool-II and an enormous amount of too-clever-by-half scripting to provide automatic load balancing across a Google CloudSQL Postgres database and one or more read replicas in the same region.

The primary moving parts are:

PGPool itself. This runs in the pod's primary container, launched from the pgpool.sh wrapper script
A discovery script that runs in a separate container. Once a minute, this script:
- Uses the gcloud cli to list any primary dbs that match the PRIMARY_INSTANCE_PREFIX env variable as a prefix. (i.e. a prefix of metadata- matches metadata-4093504851)
- Lists all currently active read replicas of the primary database
- Uses the envtpl tool to fill out pgpool.conf.tmpl and copy the result into /etc/pgpool/pgpool.conf if it differs from what is already there.
- If replicas have been added or removed, it runs pcp_reload_config which forces the pgpool process to re-read its configuration file.
- If replicas have been added, it runs pcp_attach_node to tell pgpool to begin routing traffic to it
The pgpool prometheus exporter, which exposes a Prometheus-compatible /metrics HTTP endpoint with statistics gleaned from PGPool's internal stats commands.
The telegraf monitoring agent, configured to forward the metrics exposed by the exporter to Google Cloud Monitoring, formerly known as Stackdriver.

In general you should expect that if you add a new replica, it should start taking 1/Nth of the select query load (where N is the number of replicas) within 5 to 15 minutes (beware of stackdriver reporting lag!).

By default as we've configured it, pgpool will send all mutating statements (UPDATE/INSERT/DELETE and all statements including SELECT inside a transaction with a mutating statement) to the primary instance, and load balance (on a per-statement level) all SELECTs evenly across the available read replicas.

The exception is that any statement containing the string DO NOT LOAD BALANCE (inside a comment is okay and in fact expected) will be routed to the primary instance no matter what. This is configureable at deploy time as .pgpool.primaryRoutingQueryPatternList in values.yaml.

Upgrade Guides

Old Version	New Version	Upgrade Guide
v1.1.1	v1.1.2	link
v1.1.0	v1.1.1	link
v1.0.X	v1.1.0	link

Installing the Chart

Step One: add our Helm repository to your client:

helm repo add pgpool-cloudsql https://odenio.github.io/pgpool-cloudsql && \
helm repo update

Step Two: install the chart

export RELEASE_NAME=my-pgpool-service # a name (you will need 1 installed chart for each primary DB)
export NAMESPACE=my-k8s-namespace     # a kubernetes namespace
export CHART_VERSION=1.1.2           # a chart version: https://github.com/odenio/pgpool-cloudsql/releases
export VALUES_FILE=./my_values.yaml   # your values file

helm install \
  "${RELEASE_NAME}" \
  pgpool-cloudsql/pgpool-cloudsql \
  --atomic \
  --timeout=5m0s \
  --namespace "${NAMESPACE}" \
  --version "${CHART_VERSION}" \
  --values "${VALUES_FILE}"

Please note that there are several Helm values that you are required to fill out; a bare-minimum local my_values.yaml would look like this:

discovery:
  primaryInstancePrefix: my-cloudsql-instance-name-prefix
pgpool:
  srCheckUsername: <postgres_username>
  srCheckPassword: <postgres_password>
  healthCheckUsername: <postgres_username>
  healthCheckPassword: <postgres_password>
exporter:
  postgresUsername: <postgres_username>
  postgresPassword: <postgres_password>

See below for the full list of possible values:

Chart Values

Kubernetes deployment options

Parameter	Description	Default
`deploy.replicaCount`	Number of pod replicas to deploy	`1`
`deploy.repository`	Docker image repository of the runtime image	`odentech/pgpool-cloudsql`
`deploy.tag`	Docker image tag of the runtime image	`1.1.0`
`deploy.service.tier`	Value for the "tier" label applied to the kubernetes service	`db`
`deploy.service.additionalLabels`	Map of additional k/v string pairs to add as labels for the kubernetes service	`{}`
`deploy.affinity`	Kubernetes affinity spec applied to the deployment pods	`{}`
`deploy.tolerations`	Kubernetes tolerations spec applied to the deployment pods	`{}`
`deploy.podDisruptionBudget.maxUnavailable`	Maximum number of pods allowed to be unavailable during an update (docs)	1
`deploy.resources.pgpool`	Kubernetes resource block for the pgpool container	`{}`
`deploy.resources.discovery`	Kubernetes resource block for the discovery container.	`{}`
`deploy.resources.exporter`	Kubernetes resource block for the pgpool2_exporter container.	`{}`
`deploy.resources.telegraf`	Kubernetes resource block for the telegraf container.	`{}`

PCP options

Parameter	Description	Default
`pcp.password`	Password used for connecting to the pgpool process control interface by PCP commands. This is interpolated at startup into both the pcp.conf and pcppass files so that the discovery script can send commands to pgpool. (If unset, a random one is generated at runtime.)	`""`

Discovery options

Parameter	Description	Default
`discovery.primaryInstancePrefix`	REQUIRED Search sting used to find the primary instance ID; is fed to `gcloud sql instances list --filter name:${PRIMARY_INSTANCE_PREFIX}`. Must match only one instance.	(none)

pgpool2_exporter options

Parameter	Description	Default
`exporter.postgresUsername`	REQUIRED Username used by pgpool2_exporter to connect to pgpool with. This must be a valid postgres username in your installation.	`""`
`exporter.postgresPassword`	REQUIRED Password used with `exporter.postgresUsername`	`""`
`exporter.postgresDatabase`	Postgres database used in the connection string by `pgpool2_exporter` -- this must be a valid database name in your Postgres installation.	`postgres`
`exporter.exitOnError`	Exit the container if the exporter process exits	`false`

telegraf options

Parameter	Description	Default
`telegraf.exitOnError`	Exit the container if the telegraf process exits	`false`

pgpool options

Parameter	Description	Default
`pgpool.reservedConnections`	When this parameter is set to 1 or greater, incoming connections from clients are not accepted with error message "Sorry, too many clients already", rather than blocked if the number of current connections from clients is more than (`numInitChildren` - `reservedConnections`). (docs)	`0`
`pgpool.numInitChildren`	The number of preforked Pgpool-II server processes. (docs)	`32`
`pgpool.maxPool`	The maximum number of cached connections in each Pgpool-II child process. (docs)	`32`
`pgpool.childLifeTime`	Specifies the time in seconds to terminate a Pgpool-II child process if it remains idle. (docs)	`5min`
`pgpool.childMaxConnections`	Specifies the lifetime of a Pgpool-II child process in terms of the number of client connections it can receive. Pgpool-II will terminate the child process after it has served child_max_connections client connections and will immediately spawn a new child process to take its place. (docs)	`8192`
`pgpool.connectionLifetime`	Specifies the time in seconds to terminate the cached connections to the PostgreSQL backend. This serves as the cached connection expiration time. (docs)	`0`
`pgpool.clientIdleLimit`	Specifies the time in seconds to disconnect a client if it remains idle since the last query. (docs)	`300`
`pgpool.ignoreLeadingWhiteSpace`	When set to on, Pgpool-II ignores the white spaces at the beginning of SQL queries in load balancing. (docs)	`on`
`pgpool.primaryRoutingQueryPatternList`	Specifies a semicolon separated list of regular expressions matching SQL patterns that should always be sent to primary node. (docs)	`.DO NOT LOAD BALANCE.`
`pgpool.allowSqlComments`	When set to on, Pgpool-II ignore the SQL comments when identifying if the load balance or query cache is possible on the query. When this parameter is set to off, the SQL comments on the query could effectively prevent the query from being load balanced or cached. (docs)	`on`
`pgpool.disableLoadBalanceOnWrite`	Specify load balance behavior after write queries appear. Options are `off`, `transaction`, `trans_transaction`, `always` and `dml_adaptive`. (docs)	`transaction`
`pgpool.statementLevelLoadBalance`	When set to on, the load balancing node is decided for each read query. When set to off, load balancing node is decided at the session start time and will not be changed until the session ends. (docs)	`on`
`pgpool.logMinMessages`	Controls which minimum message levels are emitted to log. Valid values are `DEBUG5`, `DEBUG4`, `DEBUG3`, `DEBUG2`, `DEBUG1`, `INFO`, `NOTICE`, `WARNING`, `ERROR`, `LOG`, `FATAL`, and `PANIC`. (docs)	`ERROR`
`pgpool.logErrorVerbosity`	Controls the amount of detail emitted for each message that is logged. Valid values are `TERSE`, `DEFAULT`, and `VERBOSE`, each adding more fields to displayed messages. `TERSE` excludes the logging of `DETAIL`, `HINT`, `QUERY` and `CONTEXT` error information. (docs)	`TERSE`
`pgpool.srCheckUsername`	REQUIRED Specifies the PostgreSQL user name to perform streaming replication check. The user must have LOGIN privilege and exist on all the PostgreSQL backends. (docs)	`""`
`pgpool.srCheckPassword`	Specifies the password of the `srCheckUsername` PostgreSQL user to perform the streaming replication checks. Use '' (empty string) if the user does not requires a password. (docs)	`""`
`pgpool.srCheckDatabase`	Specifies the database to perform streaming replication delay checks. (docs)	`postgres`
`pgpool.healthCheckUsername`	REQUIRED Specifies the PostgreSQL user name to perform health check. The same user must exist in all the PostgreSQL backends. (docs)	`""`
`pgpool.healthCheckPassword`	Specifies the password for the PostgreSQL user name configured in health_check_user to perform health check. The user and password must be same in all the PostgreSQL backends. (docs)	`""`
`pgpool.healthCheckDatabase`	Specifies the PostgreSQL database name to perform health check. (docs)	`postgres`
`pgpool.coredumpSizeLimit`	Size limit in blocks of core files that pgpool is allowed to emit if a worker crashes; this is fed to `ulimit -c` in the wrapper script, so valid values are any integer or `"unlimited"`.	`"0"`

Monitoring with Google Cloud Monitoring (AKA Stackdriver)

As configured, pgpool-cloudsql exports the following Google Cloud Monitoring metricDescriptors with the gke_container resource type and all resource labels automatically filled in.

An example Stackdriver dashboard definition can be found in monitoring/dashboard.json.

The full list of metricDescriptor definitions is in monitoring/metrics.json.

Full metric descriptor list

All metricDescriptors are created under the custom.googleapis.com/telegraf/ prefix.

Metric Descriptor	List of Metric Labels
`pgpool2_backend_by_process_total/gauge`	`pool_pid, host, url`
`pgpool2_backend_by_process_used/gauge`	`backend_id, pool_id, username, pool_pid, host, url, database`
`pgpool2_backend_by_process_used_ratio/gauge`	`pool_pid, host, url`
`pgpool2_backend_total/gauge`	`host, url`
`pgpool2_backend_used/gauge`	`host, url`
`pgpool2_backend_used_ratio/gauge`	`host, url`
`pgpool2_frontend_total/gauge`	`host, url`
`pgpool2_frontend_used/gauge`	`username, host, url, database`
`pgpool2_frontend_used_ratio/gauge`	`host, url`
`pgpool2_last_scrape_duration_seconds/gauge`	`host, url`
`pgpool2_last_scrape_error/gauge`	`host, url`
`pgpool2_pool_backend_stats_ddl_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_delete_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_error_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_fatal_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_insert_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_other_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_panic_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_select_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_status/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_backend_stats_update_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_cache_cache_hit_ratio/gauge`	`host, url`
`pgpool2_pool_cache_free_cache_entries_size/gauge`	`host, url`
`pgpool2_pool_cache_num_cache_entries/gauge`	`host, url`
`pgpool2_pool_cache_num_hash_entries/gauge`	`host, url`
`pgpool2_pool_cache_used_cache_entries_size/gauge`	`host, url`
`pgpool2_pool_cache_used_hash_entries/gauge`	`host, url`
`pgpool2_pool_health_check_stats_average_duration/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_average_retry_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_fail_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_max_duration/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_max_retry_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_min_duration/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_retry_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_skip_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_status/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_success_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_health_check_stats_total_count/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_nodes_replication_delay/gauge`	`role, hostname, host, url, port`
`pgpool2_pool_nodes_select_cnt/counter`	`role, hostname, host, url, port`
`pgpool2_pool_nodes_status/gauge`	`role, hostname, host, url, port`
`pgpool2_scrapes_total/counter`	`host, url`
`pgpool2_up/gauge`	`host, url`

background info: maybe the real friends were all the yaks we shaved along the way

A billion fussy details stood in between us and the simple-seeming state of affairs described above, and it's possible you may have to debug them in anger, so here are a few of them:

inter-container authentication

The updater container uses the pcp cli commands to talk to pgpool over its tcp control port (9898/tcp). The PCP protocol requires password auth; this is set up in the pcp.conf file that is in the shared volume mounted as /etc/pgpool in both containers. A random PCP password is set up at runtime if you do not set .pcp.password in values.yaml.

If you need to run any of the pcp_* commands yourself for interactive debugging, you will need to use the -h localhost -w flags. The -w flag forces the commands to use the /root/.pcppass file as the auth source so that the updater script (and you) can issue pcp commands without typing the password.

Note: the pcp.conf file is generated at runtime from the template.

config reloads don't do what you think

Adding a new replica to pgpool.conf and running pcp_reload_config does not commence sending traffic to that node! It only makes pgpool aware that the node exists -- you have to then run the pcp_attach_node command to tell pgpool that the node is able to receive traffic. The discovery script does this for you automatically.

node ID to hostname/ip bindings must be stable

Pgpool gets extremely unhappy if the hostname or IP address of a node changes out from underneath it; we've addressed this by adding an extremely sleazy persistence model: for every IP address we observe, we create a new node ID, and node IDs are never recycled for the lifetime of the pod.

The pgpool documentation is silent on the question of how many backend nodes it is capable of monitoring: if you delete and then re-create a large number of replicas for some reason, it might be advisable to slowly roll the pgpool pods to make sure they start from scratch.

Oh god, Stackdriver

Getting useful stackdriver metrics out of pgpool was easily 60% of the process of pulling this project together. By default, the only statistics interface that pgpool offers is a series of "SQL-like" commands that are run inside a psql session. Importantly, the tables returned by these commands are not actual tables and cannot be queried with SELECT or JOIN. (You can add the ability to do that with the pgpool_adm extension but of course there is no way to install that on a CloudSQL instance.)

But there was some good news: the pgpool2_exporter is a standalone binary that scrapes and parses the data returned by the "sql-like" commands and exports it as a prometheus-compatible /metrics endpoint.

Oh god, Telegraf

Lastly, it's worth calling out the telegraf config file; in order to correctly fill in the required attributes of a stackdriver gke_container resource, we run telegraf under a wrapper script that queries the GCP instance metadata API in order to fill out the relevant environment variables.

atbore-phx / pgpool-cloudsql Goto Github PK

pgpool-cloudsql's Introduction