Changelog

The max_worker_processes parameter's default value has been updated from 20 to 100 so that more processes can run by default, for example to run in the background for logical replication. This parameter also becomes configurable in the configuration paramaters API.

Access groups can be created with a new "Insight View" cluster permission group. This lets members view a cluster's database insights (e.g. cache rates, bloat, long running queries) even in cases where they otherwise have few privileges on the cluster.

Logical replication can now be setup for Warehouse clusters. This allows for data to be synced from a source cluster to a warehouse cluster. You can find this option in the "Data Sync" section of the Warehouse cluster menu.

Clusters now show their physical zone in the "Overview" section of the Dashboard. Zones are the underlying availability zone for AWS or Azure, or zone for GCP.

Access groups can be created with a new "Saved Query Manage" cluster permission group. This is similar to the existing "Saved Query View", but allows full access to a team's saved queries, saved query folders, and saved query links.

Crunchy Bridge is now available in the Tel Aviv, Israel (il-central-1) region for AWS.

Backup retention is now customizable in the Dashboard Cluster > Backups page. Adjust the number of daily, weekly, monthly, or, yearly additional backups. Each additional backup will be reflected in next month's invoice for a fixed rate based on storage size, but 10 daily backups are included with your plan.

A "technical email" can now to be set in Dashboard under team settings. A team's technical email is an additional non-team-member email address that will receive all technical notifications (e.g. when an autoresize is triggered or when a failover occurs) for the team the same way that admins and managers currently do.

Replicas can now be restarted from the replica settings page.

A live cluster log stream can now be viewed in the "Live Log" page.

An "Enforce MFA" option may be enabled under team settings. If enabled, all team members must have MFA (multi-factor authentication) enabled to perform any actions on the team. Users without MFA may still be members of teams with "Enforce MFA" enabled, but they won't have access to any capabilities on them until they set up a multi-factor credential on their account.

Region New Zealand North (Auckland) is now available for Microsoft Azure.

New teams are no longer assigned a "member" role/access group by default. One can be created by going to Access Groups under team settings and creating a group with the "Postgres Role" permission group. Existing teams will continue to have an existing "member" role/access group for backwards compatibility.

PgBouncer metrics are now available on the metrics page for clusters where pgBouncer is enabled, including number of client/server connections in use, bytes send/received, and average query/wait time. The new metrics will only appear if the crunchy_pooler extension has been enabled, and on servers that have been created or refreshed since January 16th, 2025.

We have disabled the high availability (HA) feature on hobby-tier clusters. Hobby instances are not intended for production purposes. If you need high availability you may resize your cluster to the standard or memory tier.

The pg_incremental extension is now available for new clusters running Postgres 16 and 17.

This extension is used in conjunction with pg_cron for running incremental batch processing jobs. See the Crunchy blog post for more information and sample use cases for pg_incremental.

The event list API endpoint supports a new delay parameter that can be used by poll loops for a stronger guarantee that events won't be missed from transactions committing out of order. We recommend using a value of 10s for a reasonable compromise between timeliness of event delivery and protection against outlying long transactions.

With Postgres 17 released in September, and 17.2 now available with a number of CVE patches, bug fixes, and improvements, we've made it the default major version for newly created clusters.

The bridge CLI tool (cb) v3.6.2 has been released. See official release for more information.

Crunchy Bridge now sends an email notification when a new cluster finishes setup and becomes ready for the first time. This behavior can be disabled from notification settings in Dashboard.

An audit log event (cluster.ready) is also produced.

Crunchy Bridge now takes two incremental backups per day and a full backup once a week. This allows for more frequent backups that are less resource intensive, and faster recovery. The change is only enabled on new clusters and existing ones that are upgraded or refreshed, but will eventually be rolled out to all clusters.

Postgres 17.2, 16.6, 15.10, and 14.15 are now available, containing fixes for four security vulnerabilities.

See the official release notes for more details.

We have released a new product offering for Crunchy Data Warehouse. Crunchy Data Warehouse replaces the earlier Crunchy Bridge for Analytics with an expanded feature set and built in managed object storage. Crunchy Data Warehouse has full Iceberg support for read, write, and update. Crunchy Data Warehouse also has extensive features to read data lake files in Parquet, JSON, and CSV.

The new clusters can be created from a dropdown toggle next to the Create Cluster options. See our documentation on Crunchy Data Warehouse. Note that warehouse has its own pricing structure and additional costs for storage.

The pg_parquet extension is now available for new clusters running Postgres 16 or 17.

See the Crunchy blog post for more information on how pg_parquet works.

The bridge CLI tool (cb) v3.6.1 has been released.

See official release for more information.

Region me-central1 is now available for GCP (Google Cloud Platform).

API endpoints to manage configuration parameters for a cluster are now available. See details in our docs.

Postgres 17 is now available. Changes include new memory management system for VACUUM, introduces JSON_TABLE for converting JSON data to a table representation, and improved logical replication (failover control, pg_upgrade - preserve replication slot for publisher & subscribers, and pg_createsubscriber - creates logical replicas from physical standbys). See the release notes for more details.

Region eu-central-2 is now available for AWS (Amazon Web Services).

We have released new options for working with spatial data in Crunchy Bridge for Analytics. Analytics clusters now support reading GeoParquet, Overture, GeoJSON, Shapefiles, CSV, and other GDAL compatible file formats. Remote object storage can be accessed via private or public S3 or public HTTPS URLs.

See our documentation on Crunchy Bridge for Analytics - Spatial on how to get started.

A new Memory-1536 plan is now available on AWS, offering 50% more memory than the previous largest plan.

The bridge CLI tool (cb) v3.6.0 has been released.

See official release for more information.

Postgres max_slot_wal_keep_size setting is now set by default to 10% of the total storage capacity, on new clusters .

Region us-west2 is now available for GCP (Google Cloud Platform).

Regions us-east1 and europe-west2 are now available for GCP (Google Cloud Platform).

A new metric view is available for network usage, showing number of bytes in and out of an instance across all its network interfaces.

Previously, saved query folders would appear for every cluster across a team, and produced odd display properties when they contained saved queries in one cluster, but were empty in another. This has now changed: like with saved queries, saved query folders are always assigned a specific cluster, and each queries view scopes folders to only its cluster. Preexisting folders that contained queries from multiple clusters have been duplicated, and saved queries reassigned to the duplicate for their specific cluster.

Crunchy Scheduler is now available for Postgres clusters. Jobs can be scheduled to run at specific times or intervals, managed through the Bridge Dashboard.

The pgx_ulid extension is now available for your Postgres clusters.

pgx_ulid adds support for creating and using ULID types in Postgres.

The pgvector extension has been updated to v0.7.0 and now supports halfvec and sparsevec types, as well as new indexing capabilities. Existing clusters can use the Refresh Instance button from the Settings tab to receive the update.

For full release notes, please review the pgvector changelog.

The pg_roaringbitmap extension has been updated to v0.5.4 and includes important fixes for users of versions 0.5.2 and 0.5.3.

For full release notes, please review the pg_roaringbitmap changelog.

To stay up to date on PostgreSQL versions offered in Crunchy Bridge, we've removed availability of Postgres 12.

See our PostgreSQL versions documentation for more information on the supported versions timeline.

The bridge CLI tool (cb) v3.5.1 has been released.

See official release for more information.

PostgreSQL 16.3, 15.7, 14.12, and 13.15 are now available, containing patches for CVE-2024-4317

For full details, please review the PostgreSQL release notes.

To encourage users to use more modern versions of Postgres, it's no longer generally possible to provision new clusters on Postgres 13. Teams that already have Postgres 13 clusters may continue to do so for the time being (for approximately the next year), but we'd encourage them to start looking into upgrading major versions as well.

When forking a cluster, saved queries are now copied into the new cluster along with the bulk of other data. These queries are copies, and updating one won't propagate back to its original. Deprovisioning the fork will also have no effect on the original cluster's saved queries.

We have released a new cluster type for Analytics. Analytics clusters support reading CSV, JSON, and Parquet files from remote object storage (e.g. S3). Clusters also enable a special vectorized and fast query engine and functions for copying files to and from the remote storage.

The new clusters can be created from a dropdown toggle next to the Create Cluster options. See our documentation on Crunchy Bridge for Analytics. Note that analytics has its own pricing structure.

The pgpodman extension that powers Container Apps has been updated to v0.3. The stop_container SQL function now has an optional second parameter to specify a timeout to wait in seconds for the container to stop. The default is 10 minutes (600 seconds). This gives control over how long to wait for a container to stop in order to help prevent a kill signal from prematurely being sent to the container.

The new timeout behavior can be obtained by running ALTER EXTENSION pgpodman UPDATE after performing a server refresh.

The pg_hint_plan extension is now available to newly created clusters. Using 'hints' in SQL comments, pg_hint_plan can adjust the execution plan of statements.

The postgres superuser has several configuration options through the ALTER SYSTEM statement. These present challenges as they do not persist on failover and are not applied to standby machines. You can now change Postgres configuration parameters in the cluster settings to work around these limitations. Direct use of ALTER SYSTEM will soon be disabled in favor of this safer configuration method.

Cluster metrics in the Dashboard support two new time ranges of 1 week and 30 days, significantly increasing the allowable lookback period. Extended periods are facilitated by a histogram-based aggregates system that makes ranging over long durations less costly to carry out.

Saved queries can now be sorted into folders to help organize them. Queries can either be top level, or stored one level deep in a folder. Nested folders are not currently supported as of this release.

Clusters can opt-in to have their statistics reset on a weekly basis with pg_stat_statements_reset() run automatically at the beginning of Sunday UTC. This helps keep query-related database insights more relevant by regularly pruning stale information. Enable the feature on a cluster's Settings page and looking for the Reset statistics weekly toggle. New clusters have it enabled by default.

You can connect a cluster to AWS PrivateLink, GCP Private Service Connect, or Azure Private Link from the cluster Networking tab. See additional details in the Private Link docs.

You can create network peering connections from inside the dashboard in the Team Settings → Networks for AWS and GCP. See additional details in the VPC peering docs.

pgBouncer 1.22.0 is now available, including improvements to prepared statements support. Existing clusters can use the Refresh Instance button from the Settings tab to get it.

PostgreSQL 16.2, 15.6, 14.11, 13.14, and 12.18 are now available, containing patches for CVE-2024-0985.

For full details, please review the PostgreSQL release notes.

The price of storage on GCP has changed from $0.10 per GB to $0.23 per GB, effective February 1st, and will apply to both existing and newly provisioned clusters. The price change in Bridge is due to an increase in disk pricing on GCP.

The bridge CLI tool (cb) v3.5.0 has been released.

See official release for more information.

The pgvector extension has been updated to v0.6.0 and now supports parallel index builds for HNSW. Existing clusters can use the Refresh Instance button from the Settings tab to receive the update.

For full release notes, please review the pgvector changelog.

The bridge CLI tool (cb) v3.4.4 has been released.

See official release for more information.

Crunchy Bridge is now available as part of the Okta integration network, which allows Okta to be configured as an OpenID Connect application for use with single sign-on. See documentation for how to configure Okta.

The bridge CLI tool (cb) v3.4.3 has been released.

See official release for more information.

The bridge CLI tool (cb) v3.4.2 has been released.

See official release for more information.

High-memory instances memory-16 through memory-864 are now available for Bridge clusters provisioned under GCP (Google Cloud Platform). Instances are based on GCP's n2-highmem-* machine types.

The kinds of notifications received from Crunchy Bridge are now configurable in Dashboard under Account Settings → Notifications, allowing users to opt out of being emailed on actions they're not interested in.

Most notifications are configurable, but some related to account security (e.g. email changed or password changed) are not.

Cluster groups are now available with support Citus Postgres extension that enables horizontal scalability with distributed storage and queries, along with columnar storage.

Create a cluster group in the Crunchy Dashboard under Team Settings → Cluster Groups, then add clusters to it from the same page.

Saved Queries in the Dashboard or API can now return up to 50,000 rows in their CSV and JSON results, up from the previous maximum of 10,000. As before, there's a limit on query results of 10 MB.

CSV or JSON must be used to get the extended result set. The maximum number of rows returned in the web UI is 1,000.

The bridge CLI tool (cb) v3.4.1 has been released.

See official release for more information.

A new standard-4 is now available to provision on AWS, coming with 4 GB of memory and 1 vCPU, with baseline IOPS of 2,500 and a maximum of 20,000.

standard-4 is available at a base price point of $70.

The pg_uuidv7 extension is now available for your Postgres clusters.

pg_uuidv7 adds support for creating and using version 7 UUIDs in Postgres.

The postgresql_anonymizer extension is now available for your Postgres clusters.

postgresql_anonymizer is an extension to mask or replace personally identifiable information (PII) or commercially sensitive data from a PostgreSQL database.

timescaledb is now available for Postgres 16 clusters.

Bridge accounts can now be created by registering a custom OpenID Connect provider, enabling access to a wider variety of identity providers and self-hosted providers. Go to OpenID Connect provider registration, verify your provider's domain, fill in client details, then complete a successful login with it to be redirected back to Bridge.

OpenID Connect providers must support the WebFinger protocol so that Bridge can verify the identity of a user with a provider before it's allowed to be added.

Passwords that are associated with both an SSO (single sign-on) provider and a password credential can now remove the latter to help better shore up the security of their account and that of teams they're members of. Passwords are considered more susceptible to attacks like credential stuffing, and the use of SSO gives administrators a faster and more definitive way of widely managing membership. Removing a password is a one-way operation. After removal, a password can't be added back.

Team administrators can go to the Members page of their teams and look for "SSO-only" badges to see which members only authenticate via SSO versus which also have a password, and may wish to ask the latter to remove their password.

Accounts can remove their password by visiting Account Settings → Authentication and looking for the "Remove Password" section. If there isn't one, no password is set.

Teams can now be configured so that they allow other accounts to join them automatically, as long as they're authenticated with the same SSO (single sign-on) provider and domain. For example, a team could be configured so that as long as a new account is authenticated through Google and have a @crunchydata.com email address, the account could join the team themselves without going through the traditional team member invite loop.

Automatic joining can be configured for a team under Team Settings → General.

Teams can be joined under Account Settings → Join Team.

With Postgres 16 available since September and 16.1 now released with fixes for three CVEs and 55 bugs (some of which affected previous versions as well), we've made it the default major version for newly created clusters.

Postgres 16.1, 15.5, 14.10, 13.13, and 12.17 are now available, containing patches for CVE-2023-5868, CVE-2023-5869, and CVE-2023-5870.

For full details, please review the PostgreSQL release notes.

The pgvector extension has been updated to v0.5.1. Existing clusters can use the Refresh Instance button from the Settings tab to receive the update.

For full release notes, please review the pgvector changelog.

Postgres' random_page_cost setting specifies the rough estimate of random reads compared to sequential ones, and helps the planner decide whether to prefer index lookups to sequential scans. Postgres' default value of 4 was originally set in 2005, a time when spinning mechanical disks were much more prolific than the SSDs generally in use today. Our testing on the three major clouds showed roughly a 5-8% cost difference between sequential and random reads, suggesting that the default random_page_cost was much too high for these environments. New Postgres servers will get a value of 1.1 instead of 4.

The list of team members for each team now shows badges indicating whether each team member has MFA (multi-factor authentication) enabled and whether their account authenticates exclusively by SSO (single sign-on) and doesn't have a password credential. This allows admins to vet the security compliance of members on their teams and reach out to those who should shore up their security posture.

Saved Queries in the Dashboard or API can now return up to 10,000 rows in their results, up from the previous maximum of 1,000.

The pgBouncer connection pooler recently announced support for a much requested feature: prepared statements. New Crunchy Bridge clusters now have pgBouncer 1.21.0 available. Existing clusters can use the Refresh Instance button from the Settings tab to get it.

It is now possible to provide your own encryption key for cluster data, otherwise known as Bring Your Own Key. See details in our docs.

The list of cluster backups now shows the version of Postgres in use at the time the backup was run.

It is now possible to view audit logs for replicas in the dashboard.

Postgres 16 is now available. Changes include improved logical replication, I/O monitoring and improved JSON functions. See the release notes for more details.

Backups and forks are now available across Postgres version upgrades, allowing safer upgrade operations and continous access to all your backups.

Redesigned the layout to improve usability of navigating around the Bridge Dashboard. Includes: - Persistent team links in top navigation bar. - Cluster dropdown that supports changing to clusters in other teams. - Some navigation moves to the left sidebar where more space is available instead of staying soley vertical.

The bridge CLI tool (cb) v3.4.0 has been released.

See official release for more information.

timescaledb is now available for your Postgres cluster.

timescaledb provides automatic partitioning of time-series data, events, and analytics.

Metric data visualizations now have better tooltips, legends, labels, and more.

Postgres 15.4, 14.9, 13.12, and 12.16 are now available, containing patches for CVE-2023-39417 and CVE-2023-39418.

For full details, please review the PostgreSQL release notes.

pglogical is now available for your Postgres cluster.

pglogical provides logical streaming replication for PostgreSQL, using a publish/subscribe model.

The metrics page now includes disk usage, which visualizes database sizes, log size, and WAL size.

The query API and saved queries in Dashboard now support Postgres range types in query results, such that they're formatted similarly to clients like psql.

Write plain text descriptions of queries and our AI-powered SQL Assistant can generate the corresponding SQL. Opt-in to share your schema for more accurate queries.

Introducing Saved Queries: Create shareable SQL queries that run against a cluster. Export Saved Queries to JSON and CSV, or embed directly into Google Sheets.

Ever wondered if your database cluster is ready for production use? There is now a production check link under 'Cluster Overview' in the Dashboard that provides detailed recommendations.

With three patch versions of Postgres 15 now released, and having been GA since October 2022, we've made it the default major version for newly created clusters.