2GB Memory limit on the graphs?

[Emilien]

Hi,
I notice this on one of my servers:

This server have installed 2x2GB memory modules:

MemTotal: 4052176 kB
MemFree: 23240 kB
Buffers: 6548 kB
Cached: 2226072 kB
SwapCached: 0 kB
Active: 1854424 kB
Inactive: 2106788 kB
SwapTotal: 3919840 kB
SwapFree: 3919676 kB
Dirty: 136 kB
Writeback: 0 kB
AnonPages: 1728600 kB
Mapped: 33108 kB
Slab: 29792 kB
SReclaimable: 17456 kB
SUnreclaim: 12336 kB
PageTables: 7684 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
WritebackTmp: 0 kB
CommitLimit: 5945928 kB
Committed_AS: 2616328 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 286100 kB
VmallocChunk: 34359451247 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB

Packages installer:

Code: Select all

NET-SNMP version:  5.4.1.1
cacti-0.8.7b

I use the Graph Template: Linux - Memory Usage

It's a bug of cacti or in net-snmp?

Best regards,

Emilien

[gandalf]

Please run snmpwalk against those OIDs (see: System Utilities -> View Poller cache, filter for host) to judge
Reinhard

[Emilien]

Uhm... I don't run snmpwalk but on Poller log I see this on the line of free memory:

Code: Select all

Script: perl /var/www/localhost/htdocs/cacti/scripts/linux_memory.pl MemFree:

It's right?

I found another host with wrong memory on the graphs, it's display 2GB but in realty have installed 1GB:

MemTotal: 1026648 kB
MemFree: 114704 kB
Buffers: 51096 kB
Cached: 313512 kB
SwapCached: 117128 kB
Active: 583796 kB
Inactive: 247800 kB
SwapTotal: 2017464 kB
SwapFree: 1757776 kB
Dirty: 92 kB
Writeback: 0 kB
AnonPages: 466428 kB
Mapped: 40136 kB
Slab: 51928 kB
SReclaimable: 23904 kB
SUnreclaim: 28024 kB
PageTables: 8808 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
WritebackTmp: 0 kB
CommitLimit: 2530788 kB
Committed_AS: 1505664 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 274384 kB
VmallocChunk: 34359463943 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB

Emilien

[IlPini]

Are the graphs for different servers the *exact* same?
In that case, please find a likely solution in graphing a linux remote host.

Forum admins... at this point, I humbly suggest some sort of FAQ or sticky treatment is in order...
(er... assuming it's not been done already... I can't find it... but, based on the number of people fooled by the names, including myself... )

[Emilien]

I set Cisco Router in all two machine.
The machine in the first post is a remote linux machine (4gb memory).
The machine on the previous post is the local linux machine where is running cacti (1gb memory)

Now I try to set "ucd/net SNMP Host" in the device of remote machine, tomorrow I'll see the results

Best regards,

Emilien

[gandalf]

Uhm... I don't run snmpwalk but on Poller log I see this on the line of free memory:

Code: Select all

Script: perl /var/www/localhost/htdocs/cacti/scripts/linux_memory.pl MemFree:

It's right?

I found another host with wrong memory on the graphs, it's display 2GB but in realty have installed 1GB:

MemTotal: 1026648 kB
MemFree: 114704 kB
Buffers: 51096 kB
Cached: 313512 kB
SwapCached: 117128 kB
Active: 583796 kB
Inactive: 247800 kB
SwapTotal: 2017464 kB
SwapFree: 1757776 kB
Dirty: 92 kB
Writeback: 0 kB
AnonPages: 466428 kB
Mapped: 40136 kB
Slab: 51928 kB
SReclaimable: 23904 kB
SUnreclaim: 28024 kB
PageTables: 8808 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
WritebackTmp: 0 kB
CommitLimit: 2530788 kB
Committed_AS: 1505664 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 274384 kB
VmallocChunk: 34359463943 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB

Emilien

Bad. That's a script for localhost only. Are you using the Host Template "Localhost" for remote targets? Please use the ucd/net snmp host template instead or the related Graph Templates.
Reinhard

[Emilien]

With ucd/net all works ok, but there is a single graph with usage cache and free memory?

Sorry for the errors, but I don't know was only for the localhost because the data in the graphs is different...

Best regards,

Emilien

[gandalf]

You may create graph layouts on your own if the default graph does not match your expectations. See the Graph template stuff at 1st link of my sig
Reinhard

[Emilien]

I resume topic because I have a similar problem with another device.
This server have 2GB of ram installed.
The Host template is:
ucd/net SNMP Host
The Graph Template is:
ucd/net - Memory Usage
See attachment, in January the memory go down to 1GB max? I haven't change the memory!
It's strange because in another device with the same config, OS and snmp version I see the correct graph with all 4GB memory.
Cacti Version 0.8.7b

Emilien

[Howie]

I think there is some other type of memory that is not accounted for - otherwise you would have a perfect flat top on the graph at 2GB, wouldn't you? Not sure what that memory is though - kernel-allocated, maybe? Actually, the memory allocated to user processes is neither buffers or cache...