very (very very) slow snmp query response

[Alond]

Hey,
I have few IRF stacks each contain 4 modules with about ~30 bridge aggregations
Im triyng to monitor thisd environment via snmp,

the issues is that Im getting very very slow snmp resoponse 

snmpwalk -v 2c -c public {irf ip} 1.3.6.1.2.1.31.1.1.1
​took:
real 17m25.470s
user 0m0.573s
sys 0m0.542s

some one know what can cause this?

snmp conf:

<10G-NG-Switch>display current-configuration | include snmp
 snmp-agent
 snmp-agent local-engineid ************
 snmp-agent community read H****
 snmp-agent community read d******
 snmp-agent community read p****
 snmp-agent sys-info location ******
 snmp-agent sys-info version all 
 snmp-agent group v3 V3authNoPriv authentication
 snmp-agent usm-user v3 snmp2user V3authNoPriv cipher authentication-mode md5 *********
 snmp-agent community-map devops context devops
<10G-NG-Switch>

this is the version:

<10G-NG-Switch>display version
HPE Comware Software, Version 7.1.045, Release 2422P03
Copyright (c) 2010-2016 Hewlett Packard Enterprise Development LP
HPE 5900AF-48XG-4QSFP+ Switch uptime is 39 weeks, 3 days, 3 hours, 6 minutes
Last reboot reason : Cold reboot

Boot image: flash:/5900_5920-cmw710-boot-r2422p03.bin
Boot image version: 7.1.045, Release 2422P03
Compiled Sep 14 2016 16:00:00
System image: flash:/5900_5920-cmw710-system-r2422p03.bin
System image version: 7.1.045, Release 2422P03
Compiled Sep 14 2016 16:00:00

Slot 1:
Uptime is 39 weeks,3 days,3 hours,6 minutes
5900AF-48XG-4QSFP+ Switch with 2 Processors
BOARD TYPE: 5900AF-48XG-4QSFP+ Switch
DRAM: 2048M bytes
FLASH: 512M bytes
PCB 1 Version: VER.B
Bootrom Version: 147
CPLD 1 Version: 004
CPLD 2 Version: 002
Release Version: HPE 5900AF-48XG-4QSFP+ Switch-2422P03
Patch Version : None
Reboot Cause : ColdReboot
[SubSlot 0] 48SFP Plus+4QSFP Plus

Slot 2:
Uptime is 39 weeks,3 days,2 hours,56 minutes
5900AF-48XG-4QSFP+ Switch with 2 Processors
BOARD TYPE: 5900AF-48XG-4QSFP+ Switch
DRAM: 2048M bytes
FLASH: 512M bytes
PCB 1 Version: VER.B
Bootrom Version: 147
CPLD 1 Version: 004
CPLD 2 Version: 002
Release Version: HPE 5900AF-48XG-4QSFP+ Switch-2422P03
Patch Version : None
Reboot Cause : ColdReboot
[SubSlot 0] 48SFP Plus+4QSFP Plus

Slot 3:
Uptime is 39 weeks,3 days,2 hours,55 minutes
5900AF-48XG-4QSFP+ Switch with 2 Processors 
BOARD TYPE: 5900AF-48XG-4QSFP+ Switch
DRAM: 2048M bytes
FLASH: 512M bytes
PCB 1 Version: VER.B
Bootrom Version: 147
CPLD 1 Version: 004
CPLD 2 Version: 002
Release Version: HPE 5900AF-48XG-4QSFP+ Switch-2422P03
Patch Version : None
Reboot Cause : ColdReboot
[SubSlot 0] 48SFP Plus+4QSFP Plus

Slot 4:
Uptime is 30 weeks,5 days,18 hours,26 minutes
5900AF-48XG-4QSFP+ Switch with 2 Processors
BOARD TYPE: 5900AF-48XG-4QSFP+ Switch
DRAM: 2048M bytes
FLASH: 512M bytes
PCB 1 Version: VER.B
Bootrom Version: 147
CPLD 1 Version: 004
CPLD 2 Version: 002
Release Version: HPE 5900AF-48XG-4QSFP+ Switch-2422P03
Patch Version : None
Reboot Cause : AutoUpdateReboot
[SubSlot 0] 48SFP Plus+4QSFP Plus

[mottesle]

I have the same issues .. on almost all our IRF comware based installation .. a snmpwalk is really slow.
When I monitor the switch using check_mk it take about 100seconds to query some ports out of a 6-member irf setup.

It seams that the "snmp" performance is really bad on comware based switches ..  on procurve / aruba bases switches the speed is much much faster .. also when they are virtualized (using vsf for example).

>> I think this is really something that should be tested + improved by HPE itself ;-))

[cgu]

hi, this is not normal numbers. What I have is about 15 seconds per switch in the stack to do complete snmp walk.

check if cpu usage is normal on every unit of the stack : dis cpu-usage slot <1-4>