Hi,
I'm quite new to cacti,
I am planning to use cacti to monitor some rather big fileservers (25-50 TB) and I get very wrong numbers.
Fileserver is a centos6 machine with a 25 terabyte samba share.
cacti is running v 0.8.8h on a centos 7 box.
cacti is reporting : used space - /mnt/data: total current 1022.01 G, used current 2.72 T
where df -h is reporting: 19TB used, 25TB total, 6.3TB available.
which is... very different.
... any advice ?
thanks,
Fab
false reporting for big disks ? (25 TB)
Moderators: Developers, Moderators
Re: false reporting for big disks ? (25 TB)
https://bugzilla.redhat.com/show_bug.cgi?id=654384
http://rhn.redhat.com/errata/RHBA-2011-1076.html
This is a problem with Net-SNMP on your Linux box. RFC says that Disk OIDs are to use 32 Bit counters. It was written well before any TB size drives were ever created.
The total size of the size is calculated as
size = hrStorageSize x hrStorageAllocationUnits
used space = hrStorageUsed x hrStorageAllocationUnits
OIDs:
hrStorageAllocationUnits = 1.3.6.1.2.1.25.2.3.1.4
hrStorageSize = 1.3.6.1.2.1.25.2.3.1.5
hrStorageUsed = 1.3.6.1.2.1.25.2.3.1.6
Net-SNMP is overflowing the hrStorageSize counter, which is then giving the low numbers.
http://rhn.redhat.com/errata/RHBA-2011-1076.html
This is a problem with Net-SNMP on your Linux box. RFC says that Disk OIDs are to use 32 Bit counters. It was written well before any TB size drives were ever created.
The total size of the size is calculated as
size = hrStorageSize x hrStorageAllocationUnits
used space = hrStorageUsed x hrStorageAllocationUnits
OIDs:
hrStorageAllocationUnits = 1.3.6.1.2.1.25.2.3.1.4
hrStorageSize = 1.3.6.1.2.1.25.2.3.1.5
hrStorageUsed = 1.3.6.1.2.1.25.2.3.1.6
Net-SNMP is overflowing the hrStorageSize counter, which is then giving the low numbers.
Re: false reporting for big disks ? (25 TB)
Hi, thank you for your answer !
I was guessing that it could be something like that...
but what do you mean by "well before any TB size drives were ever created" ? Terabytes disks are available for a while,
I can clearly remember my first install of a Terabyte-sized disk array and it was definitively during the 20th century !
Should I switch to something else than snmpd to monitor a (quite modest) 100-150 TB cabinet of file servers ?
cheers,
Fab
I was guessing that it could be something like that...
but what do you mean by "well before any TB size drives were ever created" ? Terabytes disks are available for a while,
I can clearly remember my first install of a Terabyte-sized disk array and it was definitively during the 20th century !
Should I switch to something else than snmpd to monitor a (quite modest) 100-150 TB cabinet of file servers ?
cheers,
Fab
Re: false reporting for big disks ? (25 TB)
The RFC was published in 2000. The first 1 TB hard drive was released in 2007. Granted, some people had arrays of smaller disks that equaled over 1 TB before that, but it wasn't the norm.
What version of Net-SNMP does that Centos 6 machine have? Seems that net-snmp fixed this bug sometime in 5.7
https://sourceforge.net/p/net-snmp/bugs/2294/
What version of Net-SNMP does that Centos 6 machine have? Seems that net-snmp fixed this bug sometime in 5.7
https://sourceforge.net/p/net-snmp/bugs/2294/
Re: false reporting for big disks ? (25 TB)
Hello,
Is it a run on a 32 bit system ?
If so you may hit word wrap on values after [2^32] -1
--Glenn
https://en.wikipedia.org/wiki/32-bit
Is it a run on a 32 bit system ?
If so you may hit word wrap on values after [2^32] -1
--Glenn
https://en.wikipedia.org/wiki/32-bit
Who is online
Users browsing this forum: No registered users and 0 guests