ASR 1K Memory

[prima.ramadhan]

Hi,

I have a question regarding ASR1K memory. I usually use "show proc mem" to see how much memory is used by the router. Below is the output from my ASR1K

ASR-1K#sh proc mem

Processor Pool Total:  724922940 Used:  258733288 Free:  466189652

lsmpi_io Pool Total:    6295088 Used:    6294116 Free:        972

PID TTY  Allocated      Freed    Holding    Getbufs    Retbufs Process

   0   0  208666732   24488932  158210132      43855  172941389 *Init*         

   0   0          0  768105176          0       1844       1844 *Sched*        

   0   0 1270667772 1436386948   19106888      14278      13611 *Dead*         

   0   0          0          0     589992          0          0 *MallocLite*   

   1   0     585964          0     603160          0          0 Chunk Manager  

   2   0        236        236      11196          0          0 Load Meter     

   3   0 1818002528 1812700144      30216   44975408   44975408 IP SNMP        

   4   0          0          0      17196          0          0 Retransmission o

   5   0      35016          0      17196          0          0 IPC ISSU Dispatc

I have few questions for above output:

1. There are two memory, Processor Pool and lsmpi_io pool, can anyone tell the difference?

2, Why is my lsmpi_io pool memory is used up? I only have 972 byte left. and what can I do to free up the memory?

Another thing is there is a command "show platform software status control-processor brief". Below is the output on my router:

ASR-1K#show platform software status control-processor brief

Load Average

Slot  Status  1-Min  5-Min 15-Min

  RP0 Healthy   0.00   0.09   0.15

ESP0 Healthy   0.00   0.00   0.00

SIP0 Healthy   0.00   0.00   0.00

Memory (kB)

Slot  Status    Total     Used (Pct)     Free (Pct) Committed (Pct)

  RP0 Healthy  1824960  1775944 (97%)    49016 (3%)   1476032 (81%)

ESP0 Healthy  3887632   607448 (16%)  3280184 (84%)   2557700 (66%)

SIP0 Healthy   477712   409444 (86%)    68268 (14%)    339324 (71%)

CPU Utilization

Slot  CPU   User System   Nice   Idle    IRQ   SIRQ IOwait

  RP0    0   1.30   1.20   0.00  97.49   0.00   0.00   0.00

ESP0    0   0.30   0.40   0.00  99.30   0.00   0.00   0.00

SIP0    0   0.50   0.30   0.00  99.10   0.00   0.10   0.00

Few questions for above output:

1. why is the memory usage different (the memory part). with this command it shows that RP0 memory is already used 97%, while using the previous command, the value is different.

2. how can I see which process caused this? because I think I will want to shut that process.

I have read this document (http://www.cisco.com/en/US/products/ps9343/products_tech_note09186a0080af252a.shtml)

, but I still do not understand. It does not explain the output one by one.

Regards,

Prima

[prima.ramadhan]

Anyone?

[rojesara.prashant]

Did you get any solution ? Is it a bug ?

Although in show platform software status control-processor brief RP0 usage shows 97% percent its showing in healthy state. We are having the same issue.

[Georg Pauwen]

Hello,

this is only a partial answer (excerpt from the document linked below):

2, Why is my lsmpi_io pool memory is used up? I only have 972 byte left. and what can I do to free up the memory?

On the ASR1K platform, the lsmpi_io pool has little free memory “generally less than 1000 bytes“ which is normal. Cisco recommends that you disable monitoring of the LSMPI pool by the network management applications in order to avoid false alarms.

https://www.cisco.com/c/en/us/support/docs/routers/asr-1000-series-aggregation-services-routers/116777-technote-product-00.html

[rojesara.prashant]

I was having similar issue and as per Cisco as long as committed memory is below threshold nothing to worry about.

The threshold values vary from module to module. Generally speaking, the range is:

healthy: Under 90%-95%

Warning: 90% -95%

Critical: 95% - 100%

Response in our case.

==================

The high used memory in “show platform software status control” does not mean there’s a problem with the memory. The kernel will use free memory space to grow it’s cache which will lead to the high Used memory utilization over time. These cache memory can be claimed by any kernel process when it’s needed.

Actually the system is tracking the “committed” memory utilization on memory health. The commited memory is an estimate of how much memory you need to make sure the system does not run out of memory for current workload. The kernel will over commits the memory like Linux.

At the moment although the committed memory is at 89%, it is still below the threshold hence it reported healthy memory state in RP.

 =============