Pg-spot-operator

Latest version: v0.9.47

Safety actively analyzes 723177 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 2 of 5

0.9.31

-------------------
- Release: version 0.9.31 🚀 [Kaarel Moppel]
- Better max_parallel_workers tuning. [Kaarel Moppel]
- Respect --connstr-format=ansible for --connstr-output-path. [Kaarel
Moppel]

Was storing PG connstr in normal non-vm-only mode
- Fix local file connstr integration via --connstr-output-path. [Kaarel
Moppel]
- README updates on commerical edition / features. [Kaarel Moppel]

0.9.30

-------------------
- Release: version 0.9.30 🚀 [Kaarel Moppel]
- AWS AMIs search - ignore "daily" AMIs. [Kaarel Moppel]

As not really test according to google
- Example manifest - remove deprecated comment about explicit NICs.
[Kaarel Moppel]
- Don't create explicit NICs anymore. [Kaarel Moppel]

This effectively disables persitent private IP addresses...but not a
huge loss anyways, as this disabled moving between AZs once launced
- Fix new CLI flags handling for non-default values. [Kaarel Moppel]

Logically false string were not set to empty
- Wait less on teardown befor deleting Elastic IPs. [Kaarel Moppel]

60s > 10s
- Change CLI input flags logic - all boolean flags accepting values now.
[Kaarel Moppel]

--check-price > --check-price=yes/true/on.

To a) be better aligned with K8s env params b) to be able to override
e.g. public IP on the CLI without using ENV
- CLI: add --ip-floating flag. [Kaarel Moppel]

To control Elastic IP creation
- K8s README - document how to create a K8s service for an instance.
[Kaarel Moppel]

To use the external VM transparently from within the K8s cluster
- Helm - set pullPolicy to IfNotPresent. [Kaarel Moppel]

Was "Always" - but not using tags yet, meaning could get a breaking API
change with Always
- Merge pull request 119 from pg-spot-ops/dont-retry-vm-create-on-
quota-breach. [Kaarel Moppel]

Dont retry vm create on Spot vCPU quota breach
- Show a nicer error when Spot quota exceeded. [Kaarel Moppel]

As quite a common case probably for first-time users
- VM creation - don't re-try during one loop in case Spot quota reached.
[Kaarel Moppel]
- Merge pull request 118 from pg-spot-ops/os-disk-size. [Kaarel Moppel]

New HW attribute: --os-disk-size with 20GB default
- New HW attribute: --os-disk-size with 20GB default. [Kaarel Moppel]
- Run tests on only Python 3.12. [Kaarel Moppel]

Faster feedback loop, linting should catch most of version differences

0.9.21

-------------------
- Release: version 0.9.21 🚀 [Kaarel Moppel]
- Minor correct last. [Kaarel Moppel]
- Add jumpy CPU perf to common issues README. [Kaarel Moppel]
- Env input naming - get rid of the PGSO prefix. [Kaarel Moppel]

Should be fine without it also. A bit unwieldy still
- Rename --connstr-output-only to just --connstr-only. [Kaarel Moppel]
- Merge pull request 117 from pg-spot-ops/connstr-local-file-storage.
[Kaarel Moppel]

New integration option: connect string storing into a local engine node file
- Don't overwrite the --connstr-output-path file if no changes. [Kaarel
Moppel]

to the connect string
- A new app integration flag: --connstr-output-path. [Kaarel Moppel]

To stores the PG / VM connect string in plain text on the engine node,
for easy readout by apps

0.9.20

-------------------
- Release: version 0.9.20 🚀 [Kaarel Moppel]
- Add a README on Common Issues. [Kaarel Moppel]

Like fixing the Spot Quotas and enabling non-activated regions
- Merge pull request 116 from pg-spot-ops/tuning-profiles-overhaul.
[Kaarel Moppel]

Tuning profiles overhaul, add "throwaway" profile. Remove standalone tuning profile files
- Add new tuning profile - throwaway. [Kaarel Moppel]

Plus some README and refactor
- Implement non-default tuning profiles also in the new way. [Kaarel
Moppel]
- Refactor Postgres config tuning. [Kaarel Moppel]

Replace standalone tuning scripts with normal Python code.
To avoid path issues and provide better testability
- Update README.md. [Evans Akai Bekoe]
- Minor logging output improvements. [Kaarel Moppel]
- Don't show the full values for detected manifest by default. [Kaarel
Moppel]

Show keys only in normal mode, not to leak any credentials and such
to the console. Still show in --debug mode
- Docs - various clarifications. [Kaarel Moppel]
- Print out the input params / manifest only in --debug mode. [Kaarel
Moppel]

Instead of --verbose. Better for security
- Tuning profiles - int rounding for max_parallel_workers. [Kaarel
Moppel]

and max_parallel_maintenance_workers
- Tune postgresql.conf according to the exact HW created. [Kaarel
Moppel]

Previously we did it only when explicit instance types were provided
and fell back to min HW reqs, which are normally smaller than what
was actually resolved.

https://github.com/pg-spot-ops/pg-spot-operator/issues/107
- README_aws_cli_basics.md - document how to list eviction events.
[Kaarel Moppel]

0.9.15

-------------------
- Release: version 0.9.15 🚀 [Kaarel Moppel]
- Merge pull request 115 from pg-spot-ops/persistent-vm-option. [Kaarel
Moppel]

Allow running on persistent VMs via the new --persistent-vms flag
- Better logging when --persistent-vms set. [Kaarel Moppel]

Selection strategies only relevant for Spot VMs
- README document the new --persistent-vms flag. [Kaarel Moppel]
- Minor check price output table format for non-spots. [Kaarel Moppel]
- Improved ondemand sorting. [Kaarel Moppel]

Make selection strategies spot/non-spot aware and use the according
price when sorting. If using ondemand VMs redirect all advanced sort
strategies to "cheapest"
- Sort by ondemand price when --persistent-vms set. [Kaarel Moppel]
- Rename --ondemand to --persistent-vms. [Kaarel Moppel]

Should be more universally understood as ondemand is an AWS concept
- Add new --ondemand flag to run on non-Spot VMs. [Kaarel Moppel]

Add flag and propagate to run_instances
- Add a new --debug flag. [Kaarel Moppel]

To not clean up Ansible run config and logs, for troubleshooting and
forensics
- Merge pull request 113 from pg-spot-ops/list-instances-improvements.
[Kaarel Moppel]

List instances improvements - pricing info + human readable uptime
- Hint at --list-instances in the main README. [Kaarel Moppel]
- Declare the humanize Py dependency. [Kaarel Moppel]
- Replace LaunchTime with Uptime in --list-instances. [Kaarel Moppel]

For a nicer experience
- Remove double-printing of a region's instance count in --verbose mode.
[Kaarel Moppel]
- Add current Spot price to --list-instances output. [Kaarel Moppel]
- List instances: make per region scanning messages debug level. [Kaarel
Moppel]

Show wnly the output table in normal / non-verbose mode
- Merge pull request 110 from pg-spot-ops/s3-connstr-storage. [Kaarel
Moppel]

New app integration feature - connect string info pushing to S3
- Update README_env_options.md on S3 push flags. [Kaarel Moppel]
- Document the new connstr S3 pushing feature. [Kaarel Moppel]
- Better debug info around s3 connstr storage. [Kaarel Moppel]
- Rename --connstr-bucket-filename to --connstr-bucket-key. [Kaarel
Moppel]

As more correct
- Connect info written to s3 after a success loop. [Kaarel Moppel]

If --connstr-bucket and --connstr-bucket-filename set
- Dummy connstr s3 writing. [Kaarel Moppel]
- Lay groundwork for new s3 connect string integration possibility.
[Kaarel Moppel]

New manifest section: integrations

0.9.12

-------------------
- Release: version 0.9.12 🚀 [Kaarel Moppel]
- Show Spot and On-demand prices without decimals if > $100. [Kaarel
Moppel]
- Set a "Name" tag on VM create automatically if not set by the user.
[Kaarel Moppel]

In the form of '$instance-name [pg-spot-operator]'.

To provide better recognition in the AWS Web Console.
- Merge pull request 109 from pg-spot-ops/list-instances. [Kaarel
Moppel]

New CLI flag: --list-instances
- Docement the new --list-instances option. [Kaarel Moppel]
- Add some more fields to the --list-instances output table. [Kaarel
Moppel]
- New CLI flag: --list-instances. [Kaarel Moppel]

Show key attributes from boto3.describe_instances
- Expose --storage-speed-class manifest attribute via the CLI. [Kaarel
Moppel]

As sometimes actually want HDD instances for testing
- Merge pull request 108 from pg-spot-ops/ansible-connstr-output.
[Kaarel Moppel]

Add an option to get Ansible inventory suitable output in --connstr-output-only mode
- Don't try to download Ansible files when --vm-only set. [Kaarel
Moppel]

As not needed
- README - mention that one can also use the operator for non-Postgres
tasks. [Kaarel Moppel]

And generate an Ansible inventory file plus just run custom playbooks
- Rename --connstr-output-format to --connstr-format. [Kaarel Moppel]
- Add new --connstr-output-format CLI flag. [Kaarel Moppel]

To get an Ansible compatible inventory string for easy custom setups,
where --connstr-output-format=ansible given
- README update - mention idling server estimates. [Kaarel Moppel]
- README - a more complete PyPI example + mention brute-force
protectoion. [Kaarel Moppel]
- Merge pull request 105 from pg-spot-ops/cli-list-strategies-flag.
[Kaarel Moppel]

A new CLI flag: --list-strategies
- Turn instance selection strategies into string constants. [Kaarel
Moppel]
- Bail on invalid --selection-strategy input and list available
strategies. [Kaarel Moppel]
- Add new --list-strategies CLI flag. [Kaarel Moppel]

To explain available strategies. Also add a hint to this new flag in the
check price output

Page 2 of 5

© 2025 Safety CLI Cybersecurity Inc. All Rights Reserved.