Azure CLI Monitor Metrics

[Phoenax]

Hi!

I'm playing around with the Azure CLI monitor command, to try and retrieve some values from certain metrics.

In this, case, I want the "NormalizedRUConsumption" from a CosmosDB container/collection, within a certain timeframe.
My issue here is:

• If I don't specifiy "--end-date", it doesn't default to "Now" as the end of the timeframe. Trying to get 1 month of values, in 1H interval. Is there any restrain like a maximum set size?

• Got the values in a 1 day time frame, and they "kinda" correspond to what I see in the Azure Portal dashboard Metrics blade, but the
  "timeStamp" doesn't quite correspond to the observed. There's a somewhat 1 hour difference between them.

I'm new to all these commands, and it's too much information that needs be known, I may be lacking some information as to how all this work/connects.

Thank you for your time!

Document Details

⚠ Do not edit this section. It is required for learn.microsoft.com ➟ GitHub issue linking.

• ID: 811a2d68-f0d9-5c71-b370-60995b271bc4
• Version Independent ID: 965ae9c0-9c39-8a44-9083-11f542d04099
• Content: az monitor
• Content Source: latest/docs-ref-autogen/monitor.yml
• Service: monitoring-and-diagnostics
• GitHub Login: @rloutlaw
• Microsoft Alias: routlaw

[yonzhan]

Thank you for opening this issue, we will look into it.

[AllyW]

Hi @Phoenax , for your issues, information related as below:

1. when --end-time is not provided but start-time does, end-time = start-time + offset
2. when --start-time is not provided but end-time does, start-time = end-time - offset
3. if neither --start-time nor --end-time is provided, then, end-time = now; start-time = now - offset
4. offset is the query-range and its default value is 1H

You can check other related parameter doc here: https://learn.microsoft.com/en-us/cli/azure/monitor/metrics?view=azure-cli-latest#az-monitor-metrics-list
Hope these helps.

[Phoenax]

First of all, thank you @AllyW for the input.

I hadn't in fact realized that was how it worked (don't remember reading it in the doc page you sent, might be worth to add it as a note)

As to my second point, that still doesn't answer the difference in timestamps of values observed via Dashboard and the ones returned via use of Azure CLI commands :/

[AllyW]

@Phoenax cli or portal does not have data, they just deliver the target data from corresponding azure service. For more info, you can add --debug at the end of cli cmd (like az monitor metrics list --debug) to see actually what time value has been passed into service side and check whether it is the same as the portal side. Tell me if you still have problems.

[Phoenax]

@Phoenax cli or portal does not have data, they just deliver the target data from corresponding azure service. For more info, you can add --debug...

Hi @AllyW .
I understand they don't hold data by themselves, just pull it from somewhere else, my issue is it's display in the Portal compared to what I get in the terminal (yes, I've tried the debug option).

Example:
For Database X, for a 24H timespan, intervals of 1H, the command:
az monitor metrics list --resource /subscriptions/SUB/resourceGroups/RG/providers/Microsoft.DocumentDB/databaseAccounts/DB_ACC --metrics "NormalizedRUConsumption" --offset 24H --interval 1h --filter "DatabaseName eq 'Database X'" --debug
returns (note that I just rerun it at 11:45 a.m. before writing this), as some of it's values:

"data": [
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-01T11:45:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-01T17:45:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-01T23:45:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-02T05:45:00+00:00",
"total": null
},

What I get from this, is by May 1st, at 11:45, it registered max Consumption as 2.

When I go into the portal, and set Metrics view to Last 24 hours, with granularity of 1H, I do see 4 peaks, but they are all displayed 2 hours after (so, at 1:30, and 7:30)

If I change offset/granularity to 1 minute, they show up at 1:15/16 and 7:15/16 in portal, and in the Command:
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-01T00:16:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-01T18:16:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-02T00:15:00+00:00",
"total": null
},
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-02T06:16:00+00:00",
"total": null
},

(PORTAL)

As you can see, there's a 2 Hour discrepancy for 1H granularity, and 1 Hour discrepancy for 1m granularity.

Are they both pulling data from the same place?
If yes, shouldn't it be the same?
What's the most "trustful" "source"/which one is "more correct"?

I need this data to work on a script that applies different RU settings based on hourly consumption/price, so, I'll be calling the methods via Python.

Hope this enough to explain my doubts.

Edit:
Just noticed, depending on the start/end time, for 1 hour intervals, these values get "assigned" to a different Interval slot. For example, specifying 17:00 as end date:
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-02T06:00:00+00:00",
"total": null
},
whilst at 17:45 would give the stated above:
{
"average": null,
"count": null,
"maximum": 2.0,
"minimum": null,
"timeStamp": "2023-05-02T05:45:00+00:00",
"total": null
},

So, yet another question, where is the line defined as to which interval the value is assigned?

In the portal it's always displayed (for 1 hour intervals) as X : 30, so, would I need to "force" it in the command by setting end-time to the current hour + 30?

Sorry for the neverending questions, I appreciate your efforts in helping me!