I. IntroductionTop, uptime, cat/proc/loadavg command Load average:4.90, 5.51, 5.77 overall meaning: number of running tasks + number of queued tasks the first number for example above 4.90 represents 1 minutes of The second number of data for example above 5.51 represents the first 5 minutes of data, the third number for example above 5.77 represents the first 15 minutes of data
two. Multi-core and multi-CPU differences1 CPU Single core system load Average:1 indicates 1 running tasks, 1 CPU dual-core system load Average:2 indicates 2 running tasks
Load capacity: 2 CPU Single core = 1 CPU Dual core, 1 CPU 4 cores = 2 CPU Dual core (the total number of cores is equal, the load capacity is similar)
three. How to reflect performance issuesAssuming the system is a 1 CPU single core
- If the load is 0.7 and is longer than 0.7 (5 minutes, 15 minutes of observation), then you need to investigate the cause to avoid further deterioration
- If the load is 1 and is longer than 1, you need to immediately identify the problem and fix it
- If the load is 5, and the long time is greater than 5, the entire system will slow down, the page cannot be loaded, and other serious problems appear
If it is multi-CPU multicore, and so on, when the load average to the total number of cores, SA (System administration) should be prompted to alarm, developers will have to locate the problem, solve the problem.
Four. Possible causes of high indicators
- High number of concurrency
- Some time consuming operations occupy the CPU (note that the cache is in place)
five. ReferencesWiki:https://en.wikipedia.org/wiki/load_ (computing)
Linux Performance Monitoring--load Average