No relevant resource is found in the selected language.

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>

Reminder

To have a better experience, please upgrade your IE browser.

upgrade

How to Adjust the Cache Capacity in the Linux OS?

Publication Date:  2012-07-18 Views:  66 Downloads:  0
Issue Description
When the application on the server generates a large number of discrete I/O requests continuously, the Linux OS prompts that resources are dramatically consumed. The vmstat command is executed to check the specific value. The idle value is 0. The sum of wa and sys values is about 80%. The S5000 is logged in to from the Web page. The cache utilization is high and stays more than 80%, which indicates that the S5000 has a large number of I/O requests to process. The top or ps command is executed to find that multiple pdflash processes occupy many CPU resources. If the phenomenon continues, the server may reset.
Alarm Information
None
Handling Process

The cache capacity of the Linux server can be adjusted. Enlarge the cache capacity to delay the exhaustion of the cache space by the dirty data and to enable the system to aggregate the discrete I/O requests, reducing the pressure of the S5000 to process the discrete I/O requests and improving read/write efficiency. The procedure is as follows:

 Step 1 Run the cat /proc/sys/vm/dirty_background_ratio command to check the percentage of the resources occupied by the current dirty data of the OS to the memory capacity. The default value is 10, which indicates that the dirty data cache occupies 10% of memory capacity by default.

 Step 2 Change the value as required. Run the vi /proc/sys/vm/dirty_background_ratio command to enlarge the current value. Usually, the maximum value is 50 which indicates that the dirty data occupies up to 50% of the memory capacity.

Root Cause
On the Linux server, when the cache space is used up by the system dirty data, the system starts a large number of pdflash processes to write the dirty data to hard disks, which occupies many the system resources. In this case, as a large number of I/O requests need to be processed, the value of the system becomes high, which causes the exhaustion of the system resources and system resetting.
Suggestions
During the actual deployment of the S5000, if a large number of discrete I/O requests are generated on the server or on the S5000, you can adjust the cache capacity according to the actual situation to relieve the I/O pressure.

END