Wireless Access

Hi Team,

We are seeing bad memory in monitoring tool for the controller.

The attachment shows the files which are present in the controller ,also the storage details.

Please suggest me which files I have to delete without affecting the production and let me know whether it requires a reboot after deleting the files.

Regards,

Mallikarjun

Type "tar crash" and see if there is a crash.tar file that is generated.

Hi CJoseph, 

I tried 'tar crush' command there is no such file exist. 

Can't we delete the default. cfg files. 

Any other way to get rid of it. 

Regards, 

Mallikarjun

There are files under /tmp that only Technical Support can delete.  You might want to open a case with them so that they can do that.

You can certainly delete .cfg files that are not your current configuration.

Thank you Joseph. 

For your quick response. 

Do we need to reboot the controller after deleting the cfg files. 

Regards, 

Mallikarjun

You do not have to reboot.  cfg files really do not consume much space so deleting them might not make a big difference.

Thank you for your help. 

Regards, 

Mallikarjun

You most likely have a process that is leaking memory, this cannot be solved by deleting flash files. What is the AOS version, and, please attached the output of "show processes sort-by memory" and "show memory debug" to this post as a starting point (the output is lengthy, ensure to have ssh terminal logging enabled first).

Hi jgoff,

PFA of logs.

Regards,

Mallikarjun

Attachment(s)

Memory debug logs.txt   82 KB 1 version

the good news is that there appears to be no memory leak or anything untoward with the memory util of all the various processes. What is evident is that, according to this:

(hostname) #show memory debug 
memory snapshot: Tue Apr  7 01:58:34 2020
=========================================
Memory (Kb): total: 5172096, free: 902656

there is just under 1G of memory "free". Actually this is not the whole story, there's about 2G of memory lying around in Inactive(file) which is a normal state of affairs for a Linux based OS (as the controller is)

==================
Memory Statistics
==================
MemTotal:        5172096 kB
DPMem:           2375680 kB
MemFree:          900032 kB  << what the CLI says
Buffers:           63488 kB
Cached:           729344 kB
SwapCached:            0 kB
Active:          1872576 kB
Inactive:        1962880 kB
Active(anon):     992192 kB
Inactive(anon):        0 kB
Active(file):     880384 kB
Inactive(file):  1962880 kB  << doesnt take into account

Unfortunately determining the free memory of a Linux based system is not a simple single value, you can read up more here. In actuality the memory shown as "Inactive(file)" is also mostly available should the system need it, some 1.9G of memory.

In more recent ArubaOS versions (I see this is a 2 year uptime controller running code from 2017.... that's a while!) we have a new variable for holding what is the estimated free memory of the system, named "MemAvailable", example shown below (albeit from a lightly loaded 6.5.4.15 controller)

==================
Memory Statistics
==================
MemTotal:        5171904 kB
DPMem:           2375680 kB
MemFree:         2690368 kB  << 
MemAvailable:    3505344 kB  << 
Buffers:           56640 kB
Cached:           814976 kB
SwapCached:            0 kB
Active:          1916672 kB
Inactive:         101504 kB
Active(anon):    1798080 kB
Inactive(anon):        0 kB
Active(file):     118592 kB
Inactive(file):   101504 kB

it has a higher value than "MemoryFree" and all CLI commands, dashboards, AMON and SNMP (etc.) all now use this value.

In summary, as long as you're not finding that memory is dropping more rapidly than usual (this is key, how has it behaved over the last day, week etc.) and noting that it's normal that you will see it slowly consume memory due to caching - then things are OK. In a controller with only 5G of memory, to have around 800-900 M free is OK, and that's with real consumed memory (not this kind of cached inactive memory).

TL;DR; By my eye, everything looks OK here, if you consider to schedule an upgrade or even just a reboot, the free mem will shoot back up and the process will slowly repeat over 2years.

Thank you Jgoff  for detailed explanation. 

Regards, 

Mallikarjun