database_observability: report health of component and collectors #2392
Conversation
cristiangreco
force-pushed
the
cristian/dbo11y-components-health
branch
from
6e00e77 to
b941a40
Compare
| @@ -127,7 +132,7 @@ func (c *QuerySample) fetchQuerySamples(ctx context.Context) error { | |||
| } | |||
|
|
|||
| if strings.HasSuffix(sampleText, "...") { | |||
| level.Info(c.logger).Log("msg", "skipping parsing truncated query", "digest", digest) | |||
| level.Debug(c.logger).Log("msg", "skipping parsing truncated query", "digest", digest) | |||
There was a problem hiding this comment.
Drive-by remove noisy info log
| if len(schemas) == 0 { | ||
| level.Info(c.logger).Log("msg", "no schema detected from information_schema.schemata") | ||
| return nil | ||
| } | ||
|
|
There was a problem hiding this comment.
Drive-by: log if no schema is detected
cristiangreco
force-pushed
the
cristian/dbo11y-components-health
branch
from
b941a40 to
dc57779
Compare
Report unhealthy in case of errors when starting up the collectors or of any collector is stopped during operations.
cristiangreco
force-pushed
the
cristian/dbo11y-components-health
branch
from
dc57779 to
2d5c5de
Compare
| Start(context.Context) error | ||
| Stopped() bool |
There was a problem hiding this comment.
In the future, you might have collectors that can be considered unhealthy but are still running. A different approach to support this would be to have a CurrentHealth function in the collector interface that returns the health object. Then you would not need the healthErr attribute anymore, you would just call CurrentHealth on all the collectors in the CurrentHealth function of the component.
There was a problem hiding this comment.
Yes that's a great point. I wanted to start simple for now, as collectors are anyway not resilient at all (they'll stop as soon as any error is hit). Agree that in the future we might want to delegate the logic to the collectors themselves.
PR Description
Report unhealthy in case of errors when starting up the collectors or of any collector is stopped during operations.
Which issue(s) this PR fixes
n.a.
Notes to the Reviewer
PR Checklist