* PT-1897 pt-stalk on MySQL 8 not collecting lock information
For version prior 8.0 pt-stalk continue using Information Schema INNODB_LOCKS and INNODB_LOCK_WAITS tables for collecting lock information.
For version 8.0 and higher pt-stalk uses tables data_locks and data_lock_waits in Performance Schema
* PT-1897 - Better MariaDB support
We cannot simply compare version number with 8.0 to identify if lock tables are in Performance Schema,
because MariaDB 10.x still have them in the Information Schema.
Therefore added additional flag, indicating which syntax we should use
* Tests for PT-1897
* PT-1897: added missed samples for tests
* PT-1897 Fixed retention test
Co-authored-by: Carlos Salguero <carlos.salguero@percona.com>
PT-2091 Docs configuration for Render builds
* Added Sphinx-material theme and customizations
* Changed Makefile to use Material theme
* Customized the theme
* Added a separate config for render builds
* PT-1052: fix for pt-summary
* Re-added fix from commit 71fae6d117 into proper place: library directory
Added tests for PT-1052 (pt-summary only)
* Fixed tests for pt-summary,
set locale for pt-summary tests, so they are not affected by user environment.
Finished tests for PT-1052
* Added fix for PT-1983 into the proper place: library file
* PT-1052: fix for pt-stalk which now includes numastat data
* PT-1897 pt-stalk on MySQL 8 not collecting lock information
For version prior 8.0 pt-stalk continue using Information Schema INNODB_LOCKS and INNODB_LOCK_WAITS tables for collecting lock information.
For version 8.0 and higher pt-stalk uses tables data_locks and data_lock_waits in Performance Schema
* PT-1897 - Better MariaDB support
We cannot simply compare version number with 8.0 to identify if lock tables are in Performance Schema,
because MariaDB 10.x still have them in the Information Schema.
Therefore added additional flag, indicating which syntax we should use
* Tests for PT-1897
* PT-1979 Improved mdb summary information
Added the MongoDB instance command line options to the output.
Also fixed the spacing in the replicaset section.
* Added new template file
* Upgraded deps due to security issues
* Work on PT-2029 started
* PT-2029: Polished Perl and Bash related instructions in CONTRIBUTE.md
* PT-2029: restructure for CONTRIBUTE.md
* PT-2029: more structure changes for CONTRIBUTE.md
* PT-2029 - added Go content to CONTRIBUTE.md and removed from README.md,
reorganized CONTRIBUTE.md: removed duplicate content, added Go and documentation related stuff
After discussion change only affects error message that is more user-friendly now.
I also added two tests that check how open_tables function works in both cases:
with number of open tables less tha 1000 and greater than 1000
This commit:
1. Moves all changes to the proper place: lib/bash/collect.sh
2. Refactors pt-stalk so it is more modular
3. Places fix for PT-1734 into the right place: lib/bash/collect.sh
4. Fixes tests for PT-1336
* PT-1336 Added file name checks and -maxdepth 1 parameter
* PT-1336 pt-stalk removes user files from the destination directory
* Added check that files to be removed were created by pt-stalk
* Added option -maxdepth 1 to find command, so the tool does not search for files in the nested directories
* Added unit tests
* This should fix PT-1812 also
The amount of fields returned for disk statistics
changed in newer Linux kernel versions to 20 and
caused `pt-diskstat` to simply not provide any output
on these systems. Updated the check in the diskstats
parsing code to proceed with 20 fields as well.
Signed-off-by: Lenz Grimmer <lenz.grimmer@percona.com>
* Updated changelog for (WIP) 3.3.2
* Updated changelog
* PT-7116 Updated MyState results
* PT-7116 Updated MyState results
* PT-7116 Updated MyState results
* PMM-7116 Return the error in MyState
Don't manage the error, just bubble it up.
The client must know there was an error, even when the error is no
'real' like in Mongos instances (there is no error because there is no
replicaset for mongos or standalone)
By setting SOURCE_DATE_EPOCH builds the build binaries can become
reproducible as a rebuild will take the old build date as input.
Motivation: https://reproducible-builds.org
* PMM-8575 Fixed test
Fixed sandbox used by go programs, updated all tests.
* PMM-8575 Fixed test
Fixed sandbox used by go programs, updated all tests.
* Removed Gopkg
* Switched to go 1.16
* Updated go.mod
* Upgraded deps
* Updated deps
* Fix for PT-1983 pt-summary missing one DIMM
This is a fix for PT-1983. dmidecode now doesn't print "Speed: UKNOWN" for
missing DIMMs. pt-summary code relies on the Speed field to be there, and when
it's missing, rows get concatenated (or rather not broken down correctly).
Adjusted the relevant code to not rely on the Speed field to be present.
* Added test for the fix for PT-1983
Due to dmidecode changes in 3.2, pt-summary needs to be able to work
with memory-related info with some missing fields in the dmidecode
output. This commit adds a test with dmidecode output modified to
include new format.
In addition, fixing test that are failing due to commit c9ec74c251
* PT-1983: updated changelog to reflect the bug fixed
1. the list of operation systems is outdated
2. not on any installation page we have the list of supported OSes
3. we have supported OSes in our lifecycle doc - why add it here?
* PMM-8489 Cluster ID for standalone server
* Updated error type
* Updated MyState to ignore errors on standalone servers
* Deleted dep files
* Revert "Deleted dep files"
This reverts commit 236303196b.
* Removed go.mod &^ updated lock
* Updated mongodb driver
* PMM-8489 Updated errors list
