cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1159
Views
5
Helpful
8
Replies

Nexus 3548P has memory status: Severe "Alert" , not "OK"

sakkaveelp1
Level 1
Level 1

Please help investigate, we found the Current memory status: Severe Alert on "sh system resource" 

N3K# show system resources
Load average: 1 minute: 0.98 5 minutes: 0.41 15 minutes: 0.33
Processes : 490 total, 1 running
CPU states : 3.50% user, 14.50% kernel, 82.00% idle
CPU0 states : 0.99% user, 8.91% kernel, 90.09% idle
CPU1 states : 7.00% user, 42.00% kernel, 51.00% idle
CPU2 states : 3.00% user, 3.00% kernel, 94.00% idle
CPU3 states : 1.98% user, 6.93% kernel, 91.08% idle
Memory usage: 4031392K total, 3362456K used, 668936K free
Current memory status: Severe Alert   <-----  not "OK"

After that we tried to " sh processes memory", please provide us with for checking the process that caused high memory

NXOS: version 7.0(3)I7(5a)

 

1 Accepted Solution

Accepted Solutions

Hello!

I would not anticipate that an increase in data plane traffic would cause the memory utilization of the switch to increase. However, an increase in control plane protocol scale (e.g. more VLANs being created and forwarded across more interfaces, more routes being advertised or received by various routing protocols, etc.) would cause a slight increase in memory utilization. However, so long as you stay within the unidimensional scale limits documented in the Cisco Nexus 3548 Switch NX-OS Verified Scalability Guide, Release 7.0(3)I7(5) document, I would not anticipate you having any memory-related issues.

The only two options to reduce memory utilization in this scenario is to either disable unused features that happen to be configured on the device, or to reduce the amount of control plane scale the switch is subject to. However, either of these scenarios will only reduce memory by a handful of percentage points, since the Nexus 3000 Average Memory Utilization document I referenced in my previous post indicates that a baseline memory utilization of 79% with a blank configuration is normal.

In other words, I do not think this memory utilization is a concern, and I think that trying to reduce the memory utilization of this switch by a few percentage points is unlikely to be an efficient use of your time. If you are receiving memory utilization alerts through some sort of NMS (Network Monitoring System, such as Solarwinds, LibreNMS, Observium, etc.) I would modify the configuration of the NMS such that the threshold for high memory utilization on these switches is higher.

Thank you!

-Christopher

View solution in original post

8 Replies 8

Christopher Hart
Cisco Employee
Cisco Employee

Hello!

This output is most likely normal and is not indicative of an issue.

As you can see from the output you provided, there is a total of 4,031,392,000 bytes of memory on this switch (~4GB), and 3,361,268,000 bytes of memory (~3.3GB) are in use. That's roughly 83.4% memory utilization. If you reference the Nexus 3000 Average Memory Utilization document, we can see that a Nexus 3548 running NX-OS 7.0(3)I2(2a) (very early in the NX-OS 7.x train of software, whereas you're running 7.0(3)I7(5a) which is relatively late in the NX-OS 7.x train of software) has a memory utilization of 79% per the output of the show system resources module all command.

switch# show system resources module all

CPU Resources:
----------------------------------------------------------
CPU utilization: Module 5 seconds 1 minute 5 minutes
-----------------------------------------------------------
1 6 8 8



-----------------------------------------------------------

Processor memory: Module Total(KB) Free(KB) % Used
-----------------------------------------------------------
1 3903900 795600 79

This output was taken from a switch in the lab with blank configuration, so output from your switch (which is likely in a production environment with non-default configurations, running multiple control plane protocols across multiple interfaces) only has slightly higher memory utilization compared to the lab switch's output.

Just to reiterate, this amount of memory utilization on Nexus 3524/3548 switches running an NX-OS software release in the 7.x, 9.2(x), or 9.3(x) trains of software is normal, so the "Severe Alert" memory status in the output of the show system resources command can be safely disregarded.

I hope this helps - thank you!

-Christopher

@Christopher Hart   Thank you for explaining in this situation, please tell us >>> If assume we have incoming more traffic example 50 GB , It can make using memory working height to critical? may be causing any traffic issue?

Could you please give us reduce utilization of memory?

Thank you very much for your help.

 

Hello!

I would not anticipate that an increase in data plane traffic would cause the memory utilization of the switch to increase. However, an increase in control plane protocol scale (e.g. more VLANs being created and forwarded across more interfaces, more routes being advertised or received by various routing protocols, etc.) would cause a slight increase in memory utilization. However, so long as you stay within the unidimensional scale limits documented in the Cisco Nexus 3548 Switch NX-OS Verified Scalability Guide, Release 7.0(3)I7(5) document, I would not anticipate you having any memory-related issues.

The only two options to reduce memory utilization in this scenario is to either disable unused features that happen to be configured on the device, or to reduce the amount of control plane scale the switch is subject to. However, either of these scenarios will only reduce memory by a handful of percentage points, since the Nexus 3000 Average Memory Utilization document I referenced in my previous post indicates that a baseline memory utilization of 79% with a blank configuration is normal.

In other words, I do not think this memory utilization is a concern, and I think that trying to reduce the memory utilization of this switch by a few percentage points is unlikely to be an efficient use of your time. If you are receiving memory utilization alerts through some sort of NMS (Network Monitoring System, such as Solarwinds, LibreNMS, Observium, etc.) I would modify the configuration of the NMS such that the threshold for high memory utilization on these switches is higher.

Thank you!

-Christopher

@Christopher Hart  

I'm very grateful. for now, I have new plan to upgrade switch. 

Thank so much for your help.

switch# show processes memory <<- please share this 

@MHM Cisco World  Please see in attached file. thank you

show system inband queuing statistics 

can share this,  

@MHM Cisco World 

 For now, I have new plan to upgrade switch. 

Thank you so much for your investigation.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco