Conversation
ea0a387 to
02d1668
Compare
02d1668 to
6ac1c16
Compare
6ac1c16 to
74a21f2
Compare
f01ff44 to
30ba2cc
Compare
Codecov Report
|
l0k0ms
left a comment
There was a problem hiding this comment.
G2G doc wise after the small nits are integrated :)
Co-Authored-By: Pierre Guceski <pierre.guceski@datadoghq.com>
Co-Authored-By: Pierre Guceski <pierre.guceski@datadoghq.com>
Co-Authored-By: Pierre Guceski <pierre.guceski@datadoghq.com>
Co-Authored-By: Pierre Guceski <pierre.guceski@datadoghq.com>
Co-Authored-By: Pierre Guceski <pierre.guceski@datadoghq.com>
therve
left a comment
There was a problem hiding this comment.
Looks great! Some minor comments, mostly a doc discrepancy.
| ) | ||
|
|
||
|
|
||
| def get_check(instance): |
There was a problem hiding this comment.
For when we have multiple test files, may want to do this now #6117 (comment)
There was a problem hiding this comment.
But as a fixture you need to set an extra param in each test function.
In that case there is a single test file, but even with multiple, we could import the helper method from somewhere and I think it's cleaner to have a single import than updating every test signature.
There was a problem hiding this comment.
by that logic, what's the point of fixtures for config?
There was a problem hiding this comment.
I think a fixture is useful if you execute code in it. You read a file maybe, or you do a deepcopy of an integration config. By that mean, it abstracts a lot of complexity in your test, you just use the result of that fixture as a parameter of your tests.
If the result of your fixture is a static function that you could define what's the point? Seems more complex (the test runs the code of the fixture, gets the result, put in in an argument of the test, then you call that argument (because the result of that fixture was a function itself)....) than simply defining the function statically.
Co-Authored-By: Ofek Lev <ofekmeister@gmail.com>
Co-Authored-By: Ofek Lev <ofekmeister@gmail.com>
Co-Authored-By: Ofek Lev <ofekmeister@gmail.com>
|
Hello, this does not seem to work with ProxySQL user specified in |
|
For users in admin-stats_credentials, the 'stats' database is named 'main'. Also, there's no way for stats users to get the proxysql version. |
|
It's just my suggestion. I think it's not secure to use admin user for monitoring and I think it is possible to use user specified in |
ProxySQL is a new integration, originally implemented here
This PR is based on this implementation but a few changes are to be noted.
To show the full diff, you can use this permalink
Implementation-wise
proxysql.backend.connectis emitted for each mysql backend that proxysql knows about. The value of the service check is based on the availability of the given backend.conf.yaml.exampleis generated automatically.Metric-wise
proxysql.query_processor_time_ms) that means the number of time spent doing a given thing are converted intotemporal_percent(i.eproxysql.query_processor_time_pct) that represent the amount of time spent in that state as a percentage of total time. This is generally easier to work with.proxysql.uptimequery_cachemetrics a rategaugeinstead ofrateeverywhere. When the agent submits a rate, the backend type for that metric is agaugeslite3_memorymetric under the nameproxysql.memory.sqlite3_memory_bytes