Start a Conversation

Solved!

Go to Solution

1 Rookie

 • 

6 Posts

34

January 7th, 2025 12:43

S4148F-ON OS 10.5 high memory usage

Hello,

we have couple of S4100 switches which all have very high memory utilization during normal operation which is reaching 80% and it is concerning. Even new S4148F-ON that are in greenfield deployment have high memory utilization 65-70% without any configuration. All switches are on 10.5.4/10.5.5

Please advise how this can be addressed or how can we review which processes are responsible for which feature? 

For example:

python/python3
dn_dot1x - if we are not using .1X
app_telemetry

or different .py scripts like arp_filter.py etc. 

G.

Moderator

 • 

9.1K Posts

January 7th, 2025 16:58

Petrovg,

 

Without having more to go on we would need more details, but i'll provide some guidance on how to address this issue and review which processes are responsible for the high memory usage.

 

Initial Steps:

Check the switch logs : Review the system logs, error logs, and debug logs to look for any patterns or errors that might indicate the cause of the high memory utilization.

 

Verify switch configuration : Ensure that the switch configuration is optimized and not causing unnecessary memory usage. Check for any unnecessary features or protocols enabled.

 

Monitor memory utilization : Use the show memory command to monitor memory utilization over time and identify any trends or spikes.

 

 

 

Process Review:

To review which processes are responsible for the high memory usage, you can use the show process command. This command displays a list of running processes, including their memory usage and CPU utilization.

 

For example:
show process | include python

 

This command will display a list of processes that include the word "python" in their name. You can use this command to identify which Python-based processes are running on the switch.

 

 

Identify Unnecessary Processes:

Review the list of running processes and identify any that are not necessary or can be disabled.

For example, if you're not using 802.1X, you can disable the dn_dot1x process.

 

 

Optimize Switch Configuration:

Check the switch configuration to ensure that:

Unnecessary features or protocols are disabled
The switch is configured for optimal performance
The memory allocation is optimized

 

 

Disable Unnecessary Features:

Disable any features or protocols that are not being used. For example:

Disable the app_telemetry process if it's not being used
Disable the arp_filter process if it's not being used

 

Upgrade to a Later Firmware Version:

If you're running firmware versions 10.5.4 or 10.5.5, consider upgrading to a later version, as newer firmware versions may include bug fixes or performance improvements that can help reduce memory utilization.

 

Additional Tips:

Regularly monitor memory utilization and adjust the switch configuration as needed

 

Consider enabling memory logging to help identify memory-related issues
If you're unable to identify the cause of high memory utilization, consider opening a support ticket with Dell or seeking assistance from a networking expert.

 

I hope this helps you address the memory utilization on your Dell S4100 switches. If you have any further questions or need additional guidance, feel free to ask.



1 Rookie

 • 

6 Posts

January 17th, 2025 14:39

Hello Chris H,

Thank you for the prompt reply. We can share more details as needed. 



Initial steps: 
the logs have been reviewed and there is nothing concerning. This behaviour is observed on production switches and on newly installed switches that have no downlinks active or any traffic being switched



Verify switch configuration: 
Basically this is the main reason why this conversation was started - how can we understand which process is used for what and which could be the unnecessary processes that we can stop?



Monitor memory utilization: We are constantly monitoring the memory utilization over the past and it is increasing slowly and with no spikes

switch1 - 67%  on 10.5.5.5
switch2 - 65%  on 10.5.5.5
switch5 - 71%  on 10.5.4.0
switch6 - 68%  on 10.5.4.0
switch7 - 68%  on 10.5.4.0
switch8 - 68%  on 10.5.4.0


Process Review:
# show processes memory

Memory Statistics of Unit 1 (kilobytes)
=========================================
Total: 4012544, CurrentUsed:3507196, CurrentFree:505348

PID         Process           Total-used
1862        base_nas          357688
1867        netconfd-pro      113220
19962       .clish            107296
26555       .clish            104352
1143        dn_sm             84824
1154        dn_xstp           83756
2338        python3           76544
1171        dn_dcbx           74968
1015        python3           70608
1168        python3           70464
806         dn_dot1x          69644
1165        dn_lacp           68684
833         dn_svc_sw         64204
1547        app_telemetry     63484 


For instance how can we disable DCB, .1X and Telemetry?
They should be off by default and we have not enabled them (config-wise) though they are running as processes



Identify Unnecessary Processes:
Even though we could potentially identify the unnecessary processes I am not able to figure out how to stop them from the cli. Stopping them from the linux shell is something that I rather not do as this will not survive system reload for example



Optimize Switch Configuration:
In general our configuration is as per best practises and already optimized.
All features that are not used are disabled (not configured) but some of them show as running processes 



Upgrade to a Later Firmware Version:
I have reviewed the but fixes and I cannot see something related to memory optimization or similar.


Additional Tips:

In general all that you are listing and saying is true but it is pretty general and doesn't tell us much.
Please, write if you advise to change or do something with exact steps and example, not just general description of what should be done


Thank you in advance !

Moderator

 • 

9.1K Posts

January 17th, 2025 15:15

Thank you. As the memory isn't maxing out and it isn't currently causing an issue, it is sounding more like a performance issue which we wouldn't be able to support. You could monitor and get the memory usage every day for a week and see how it changes and see which processes are doing it, then call in to support and see about getting it analyzed and possibly escalated.

 

 

 

No Events found!

Top