TS.MRANGE works intermittently

[slorello89]

Describe the bug

See tail end of #296 for details - Moving this out into a separate issue as #296 has been addressed and this is more of a spun-out issue.

the gist of it is that when this user is using TS.MRANGE for many time-series the user is seeing inconsistent results, sometimes they'll get back 150 time series, other times they'll get back 400 etc. . . They only get what they expect about 1/5th of the time

Have tried reproducing this in my own environment to no avail

Version

Plugin v2.1.1 @NikolaBorisov to provide details of grafana/redis version

Steps to reproduce

Unknown - need @NikolaBorisov to provide a minimal repro of the issue that's repeatable outside of their environment.

Expected behavior

Should get consistent results every time the dashboard is repopulated

Screenshots

See end of #296

Additional context

Things checked:

1. Load on frontend - queries would effectively be pulling back ~300k results (400 timeseries, over 12 hours with 60 second time buckets) this does not appear to be what they are getting back.
2. Was this happening in a clustered environment - looks like there's a chance of MRANGE using random shards in a clustered environment. But the user is not using it in a clustered environment
3. Load on the backend - Redis completes the queries in a 'reasonable' time (<100ms)

[slorello89]

@NikolaBorisov - at this point I'm out of ideas and have been unable to reproduce what you are describing in my own environment - can you please provide a minimal reproduction of the issue that I can run so I can investegate?

thanks,

[NikolaBorisov]

Im traveling next 3 weeks, give me some time and i will try to reproduce