问题

I am using the prometheus query '100 - ((node_memory_MemAvailable_bytes{job="jobname"} * 100) / node_memory_MemTotal_bytes{job="jobname"})' to get the memory utilization and it is working fine.

以上查询用于获取内存利用率，它正常工作。

The above query is giving the result for current memory utilization.

上述查询提供当前内存利用率的结果。

I need to reconstruct the query to get the memory utilization for last 1 hour. For example, if the memory utilization reaches more than 10 MB at 9.15 am and the current memory utilization is 2 MB at 10 am, now I need to check whether the memory utilization is more than 9 MB between 9 am to 10 am.

我需要重构查询，以获取过去1小时内存利用率。例如，如果内存利用率在上午9:15达到10 MB以上，并且当前内存利用率在上午10点为2 MB，现在我需要检查9点到10点之间内存利用率是否超过9 MB。

Kindly guide me how to construct the prometheus query for it, I think it is something like, '(100 - ((node_memory_MemAvailable_bytes{job="jobname"} * 100) / node_memory_MemTotal_bytes{job="jobname"}))>9[1H]'

请指导我如何构建用于此目的的Prometheus查询，我认为它类似于 '(100 - ((node_memory_MemAvailable_bytes{job="jobname"} * 100) / node_memory_MemTotal_bytes{job="jobname"}))>9[1H]'。

答案1

得分: 4

以下是您要翻译的内容：

"Probably you need SLI value answering the question: how much time the jobname used more than 90% of memory over the last hour? Then the following PromQL query should answer the question:

    ((1 - node_memory_MemAvailable_bytes{job="jobname"} /
    node_memory_MemTotal_bytes{job="jobname"}) >bool 0.9)[1h:1m]
)

The returned value will be in the range [0..1], where 0 means 0% (i.e. memory usage didn't exceed 90% during the last hour), while 1 means 100% (i.e. memory usage was higher than 90% all the time during the last hour).

The query uses the following PromQL features:

• subquery
• bool modifier for comparison operator
• avg_over_time function."

avg_over_time(
    ((1 - node_memory_MemAvailable_bytes{job="jobname"} /
    node_memory_MemTotal_bytes{job="jobname"}) >bool 0.9)[1h:1m]
)

答案2

得分: 0

100 - ((node_memory_MemAvailable_bytes{job="jobname"} * 100) / node_memory_MemTotal_bytes{job="jobname"})[30m:1m]

这将列出每1分钟的内存利用率数值。