GitLab

Change to AGPLv3

See merge request !449

L003: do not analyze multi-columns PKs, fix JSON syntax

See merge request !448

- Skip multi-column PKs: analyzing each column in them may cause a huge slowdown,
- fix JSON syntax (previously, jq was not complaining, but syntax lacked some commas)

fix: dockerfile, pghrep build

See merge request !446

feat: improved docker images storage

See merge request !445

fix: typo in error message

See merge request !444

Link fix + better GIF quality

See merge request !442

No year in copyright; README: -h vs --pg-../--ssh-..

See merge request !441

README: Ask for a gitlab star

See merge request !440

Support of PG 12 was realized not fully. It is fixed.

F004 report, which used pg_class.relhasoids column does not process correctly on Postgres 12. So this report was reworked to remove using of pg_class.relhasoids.

In the case when `sudo` requires a password on the checked server(s), data collection for the A008 and D002 reports fails.
Now it is possible to specify such a password using the environment variable `SSHSUDOPASSWORD`.

Sometimes during query analisis we neet to know query id, but it does not display in K003 report. So query id added to first column of K003 report's table.

fix: run_checkup.sh

See merge request !436

Support of pg and ssh hostnames in run_checkup.sh added

fix: README section Example of Use

See merge request !434

README: link to 1.3.0 example

See merge request !430

In A008, if Postgres uses syslog for logging, it does not make sense to analyze and show details log directory/file and mount point. Now we skip details in such cases.

Recommendations and Conclusions sections of H002 and H004 reports are fully reworked. The main focus is to make it usable for large databases, where large numbers of unused and redundant indexes need cleanup. Additionally, detailed directions on cleanup tactics are provided.

In case when options `--pdf` and/or `--html` were enabled generation of HTML and PDF files was executed twice. In this MR such behaviour was fixed.

In some cases checking of foreign indexes led to duplication indexes during collection, so as result in json we had with not consecutive rows. Now it fixed.