SNMP counter - integer overflow? - toasters

List overview All Threads
Download

newer

SNMP counter - integer overflow?

older

Re: SNMP counter - integer...

alternative to ApplianceWatch

Henrique Pantarotto

13 Sep 2000 13 Sep '00

9:24 p.m.

Hello,

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the value that I am getting from the last few days are in "negative". I believe it happened an overflow in the integer type used to hold this data, right?

Look:

[root@foo /root]# snmpget filer-e0.terra.com.br public .1.3.6.1.4.1.789.1.2.2.1.0 enterprises.789.1.2.2.1.0 = -1281161705

I know I can fix this by rebooting the filer. But I don't want to do that.

This is a F760 with Ontap 5.3.6.R2. It has a lot of traffic. ;-)

Does anyone here knows of a solution to this problem?

Thanks, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

Show replies by date

James R Grinter

14 Sep 14 Sep

7:05 a.m.

On Wed, Sep 13, 2000 at 06:24:47PM -0300, Henrique Pantarotto wrote:

...

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the value that I am getting from the last few days are in "negative". I believe it happened an overflow in the integer type used to hold this data, right?

Well, it's an INTEGER so it's allowed to be negative - the mistake made was in not making it a Counter, and there's little you can do except wait for NetApp to release new code that adds to the MIB and gives proper Counters.

(Well, you can modify your polling arrangements to specifically treat the number you get back from this area of the MIB as a +ve integer. I believe someone has done just this sort of modification for Cricket. Have a read of the Cricket mailing list archives, for the thread "[cricket-users] Negative SNMP counters (netapp)")

James.

Henrique Pantarotto

12:42 p.m.

Folks, you know what I did?

A friend here at work found out another OID:

1) the one I was using before: netapp1.sysStat.misc.miscNfsOps (1.3.6.1.4.1.789.1.2.2.1) "The total number of Server side NFS calls since the last boot."

2) the other one: netapp1.nfs.curNfs.nfsServ.nfsCalls (1.3.6.1.4.1.789.1.3.1.2.1) "The total number of NFS calls received, since the last time the statistics were cleared."

I am using the second one now. It probably has the same limitations as the first, but it can be nicely zeroed with "nfsstat -z" and it won't mess up the graphics.

All I have to do is to run that command every once in a while.

Of course: I don't see this as the "perfect" solution. I would love if netapp would do a fix for this. (if one is possible?)

Thanks all for the replies!

Regards from Brazil, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

...

-----Original Message----- From: Henrique Pantarotto [mailto:henrique@corp.terra.com.br] Sent: Wednesday, September 13, 2000 6:25 PM To: toasters@mathworks.com Cc: netapp@terra.com.br Subject: SNMP counter - integer overflow?

Hello,

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the value that I am getting from the last few days are in "negative". I believe it happened an overflow in the integer type used to hold this data, right?

Look:

[root@foo /root]# snmpget filer-e0.terra.com.br public .1.3.6.1.4.1.789.1.2.2.1.0 enterprises.789.1.2.2.1.0 = -1281161705

I know I can fix this by rebooting the filer. But I don't want to do that.

This is a F760 with Ontap 5.3.6.R2. It has a lot of traffic. ;-)

Does anyone here knows of a solution to this problem?

Thanks, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

Edward Henigin

5:24 p.m.

I'm surprised noone has pointed this out yet, but there is no problem. All you are monitoring off of this are deltas. If at time T the counter is -30000, and at time T' it's -29000,

-29000 - -30000 ======== 1000

So ops/sec are 1000 / (T' - T)

The deltas work just fine, regardless of whether the counter happens to be positive or negative.

-- On Wed, Sep 13, 2000 at 06:24:47PM -0300, Henrique Pantarotto said:

...

Hello,

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the value that I am getting from the last few days are in "negative". I believe it happened an overflow in the integer type used to hold this data, right?

Look:

[root@foo /root]# snmpget filer-e0.terra.com.br public .1.3.6.1.4.1.789.1.2.2.1.0 enterprises.789.1.2.2.1.0 = -1281161705

I know I can fix this by rebooting the filer. But I don't want to do that.

This is a F760 with Ontap 5.3.6.R2. It has a lot of traffic. ;-)

Does anyone here knows of a solution to this problem?

Thanks, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

-- Edward Henigin ... We build e-business infrastructure solutions CTO, DataFoundry.net ... www.datafoundry.net

Matthew Stier

9:06 p.m.

True, MRTG uses delta's, however the newer releases of MRTG use RRDTool, and this can have problems with negative numbers in field that are defined as COUNTER.

I solve the problem with ONTAP 5.x, by simply adding 2^31 to the returned value. This changes the range from -2^31 through 2^31 to 0 through 2^32.

Edward Henigin wrote:

...

    I'm surprised noone has pointed this out yet, but there is
no problem. All you are monitoring off of this are deltas. If at time T the counter is -30000, and at time T' it's -29000,
      -29000
    - -30000
    ========
        1000

    So ops/sec are 1000 / (T' - T)

    The deltas work just fine, regardless of whether the counter
happens to be positive or negative.
    Ed
-- On Wed, Sep 13, 2000 at 06:24:47PM -0300, Henrique Pantarotto said:

...
Hello,

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the value that I am getting from the last few days are in "negative". I believe it happened an overflow in the integer type used to hold this data, right?

Look:

[root@foo /root]# snmpget filer-e0.terra.com.br public .1.3.6.1.4.1.789.1.2.2.1.0 enterprises.789.1.2.2.1.0 = -1281161705

I know I can fix this by rebooting the filer. But I don't want to do that.

This is a F760 with Ontap 5.3.6.R2. It has a lot of traffic. ;-)

Does anyone here knows of a solution to this problem?

Thanks, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

-- Edward Henigin ... We build e-business infrastructure solutions CTO, DataFoundry.net ... www.datafoundry.net

-- Matthew Lee Stier * Fujitsu Network Communications Unix Systems Administrator | Two Blue Hill Plaza Ph: 914-731-2097 Fx: 914-731-2011 | Sixth Floor Matthew.Stier@fnc.fujitsu.com * Pearl River, NY 10965

Henrique Pantarotto

9:33 p.m.

Edward,

I am tired now and am not thinking right at this moment.. ;-), but my MRTG was all crazy, creating flat lines of monitoring.. it seemed that he wasn't liking the negative numbers he was getting.. ;-(

But it's okay now (using the other OID, which can be zeroed with nfsstat -z).

Thanks!

_______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

...

-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Edward Henigin Sent: Thursday, September 14, 2000 2:24 PM To: Henrique Pantarotto Cc: toasters@mathworks.com; netapp@terra.com.br Subject: Re: SNMP counter - integer overflow?

I'm surprised noone has pointed this out yet, but there is no problem. All you are monitoring off of this are deltas. If at time T the counter is -30000, and at time T' it's -29000,
 -29000
-30000

======== 1000

So ops/sec are 1000 / (T' - T)

The deltas work just fine, regardless of whether the counter happens to be positive or negative.

Ed

-- On Wed, Sep 13, 2000 at 06:24:47PM -0300, Henrique Pantarotto said:

...
Hello,

I am monitoring nfs ops/sec in a F760 filer using MRTG, but the

value that I

...
am getting from the last few days are in "negative". I believe

it happened

...
an overflow in the integer type used to hold this data, right?

Look:

[root@foo /root]# snmpget filer-e0.terra.com.br public .1.3.6.1.4.1.789.1.2.2.1.0 enterprises.789.1.2.2.1.0 = -1281161705

I know I can fix this by rebooting the filer. But I don't want

to do that.

...
This is a F760 with Ontap 5.3.6.R2. It has a lot of traffic. ;-)

Does anyone here knows of a solution to this problem?

Thanks, _______________________________________________ Henrique Pantarotto SysOp Site São Paulo Terra Networks Brasil S/A A Internet mais sua do que nunca Tel: (11) 5505-5728 r.316/238 ICQ: 6934285 IT: henpa henrique@corp.terra.com.br

-- Edward Henigin ... We build e-business infrastructure solutions CTO, DataFoundry.net ... www.datafoundry.net

8981

Age (days ago)

8982

Last active (days ago)

toasters@lists.teaparty.net

5 comments

4 participants

tags (0)

participants (4)

Edward Henigin
Henrique Pantarotto
James R Grinter
Matthew Stier