AS400 CPU Utilization

[anh]

Do you have a working template ?

[spoonman]

Attached is what I've been able to graph from our AS400...wish i could get more!!!!

[jcheney]

I don't suppose you could export a template, or provide the OIDs you used for those graphs...

[fmangeant]

Hi

I'm using these OIDs to graph AS/400 :

• .1.3.6.1.2.1.25.1.5.0 (hrSystemNumUsers) for logged in users
• .1.3.6.1.2.1.25.1.6.0 (hrSystemProcesses) for running processes
• .1.3.6.1.4.1.2.6.4.5.1.0 for CPU usage
• hrStorageSize and hrStorageUsed for disk usage (with index=1)
• .1.3.6.1.2.1.25.2.2.0 (hrMemorySize) for memory size, and a script which get the index of memory usage in hrStorageAllocationUnits

[MBurroughs]

Spoonman, Where did you get the templates that you used to graph the AS400? Are you using AS400 temps or are they Windows templates..

[jcheney]

Thanks fmangeant, I was able to get everything but the disk and memory usage, but it would appear that this particular machine is not returning those values. When I do an snmpwalk they don't show up, and when doing an snmpget they return a noSuchName error.

Thanks for your help!

[jcheney]

I thought that I would post the template I have. It needs work still, but it might be a place for someone to start.

[spoonman]

I wish I could get more from the As400// num jobs I haven't figured out how to use the storage OID's to get Hard drive space graphs etc...

[fmangeant]

I wish I could get more from the As400// num jobs I haven't figured out how to use the storage OID's to get Hard drive space graphs etc...

For memory, I'm using this quick and dirty script :

Code: Select all

$ cat get-as400-memory.sh

#!/bin/sh

IP=$1
COMMUNITY=$2
SNMP_VERSION=$3
MACHINE=$4

# RAM total size
RAM_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrMemorySize.0 | awk '{ print $(NF-1) }'`
RAM_SIZE=`expr $RAM_SIZE`

# RAM used
NB_ELEMENTS=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageIndex | awk '{ print $NF }' | tail -n 1`

RAM_USED=0

for i in `seq 1 $NB_ELEMENTS`; do
        TYPE_ELEMENT=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageDescr.$i | awk '{ print $NF }'`
        if [ "$TYPE_ELEMENT" != "RAM" ]; then
                continue
        fi

        BLOCK_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageAllocationUnits.$i | awk '{ print $(NF-1) }'`
        NB_BLOCKS_USED=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageUsed.$i | awk '{ print $NF }'`
        RAM_USED=`expr $NB_BLOCKS_USED \* $BLOCK_SIZE`
        TOTAL_RAM_USED=`expr $TOTAL_RAM_USED + $RAM_USED`
done

TOTAL_RAM_USED=`expr $TOTAL_RAM_USED / 1024`
RAM_FREE=`expr $RAM_SIZE - $TOTAL_RAM_USED`

echo "valeur1:$RAM_FREE valeur2:$RAM_SIZE"

[spoonman]

How do you use it??

[fmangeant]

I've created a data input method which runs a script, and gets 2 results in return.

[jtnfoley]

I had to prowl around with getif to sniff out the OIDs and changed the behavior of the awks a little, but...
This seems to work with OS/400 5.3, on Fedora Core 4 as the cacti host. I'm still trying to get everything plumbed together and actually make a graph, however...
NET-SNMP version: 5.3.1, FWIW.

Code: Select all

#!/bin/sh

IP=$1
COMMUNITY=$2
SNMP_VERSION=$3
MACHINE=$4

# RAM total size
#RAM_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrMemorySize.0 | awk '{ print $(NF-1) }'`
RAM_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP .1.3.6.1.2.1.25.2.2 | awk '{ print $(NF) }'`
RAM_SIZE=`expr $RAM_SIZE`

# RAM used
#NB_ELEMENTS=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageIndex | awk '{ print $NF }' | tail -n 1`
NB_ELEMENTS=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP .1.3.6.1.2.1.25.2.3.1.1 | awk '{ print $NF }' | tail -n 1`
RAM_USED=0

for i in `seq 1 $NB_ELEMENTS`; do
        #TYPE_ELEMENT=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageDescr.$i | awk '{ print $NF }'`
        TYPE_ELEMENT=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP .1.3.6.1.2.1.25.2.3.1.3.$i | awk '{ print $(NF-1) }'`
        if [ "$TYPE_ELEMENT" != "\"RAM" ]; then
                continue
        fi

#        BLOCK_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageAllocationUnits.$i | awk '{ print $(NF-1) }'`
        BLOCK_SIZE=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP .1.3.6.1.2.1.25.2.3.1.4.$i | awk '{ print $(NF) }'`
       # NB_BLOCKS_USED=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP hrStorageUsed.$i | awk '{ print $NF }'`
        NB_BLOCKS_USED=`snmpwalk -v $SNMP_VERSION -c $COMMUNITY $IP .1.3.6.1.2.1.25.2.3.1.6.$i | awk '{ print $NF }'`
        RAM_USED=`expr $NB_BLOCKS_USED \* $BLOCK_SIZE`
        TOTAL_RAM_USED=`expr $TOTAL_RAM_USED + $RAM_USED`
done

TOTAL_RAM_USED=`expr $TOTAL_RAM_USED / 1024`
RAM_FREE=`expr $RAM_SIZE - $TOTAL_RAM_USED`

echo "valeur1:$RAM_FREE valeur2:$RAM_SIZE"

[jtnfoley]

My D'OH! moment of the day...
Put the mib in /usr/local/share/snmp/mibs/ and changed the snmpwalk commands to include "-m all" and the (otherwise) original script now works. With "-m all" the performance is far slower than with the OIDs above...
"-m IBM-AS400-RWS-MIB.txt" errors, even with absolute path...

[Alberello]

Hello,
I am just able to monitor the number of process...

what I have to put here in OID?
I have a OS/400 V5R3M0

[dafa]

Can someone post an example of how to use the sh script to check AS/400 memory usage ?

Thanx.