[Check_mk (english)] 1.2.8b7 has weird limits on levels

Archive checkmk-en
1

Hi CMK list,

I'm trying to tune rules for Virtual Memory Warning and Alerting level.
In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on KVM hypervisors)

However I get these error in WATO:
    150.0 is too high. The maximum allowed value is 101.0
    250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

-- Hans

2

Hi Hans,

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will never get >100%. 101.0 is used to disable the threshold. If you want to use the “old” thresholds in the new linux memory check you’ve to set param “Upper levels for Total Data in relation to RAM” I think.

HTH,

Marcel

···

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um 13:13 Uhr:

Hi CMK list,

I’m trying to tune rules for Virtual Memory Warning and Alerting level.

In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on KVM hypervisors)

However I get these error in WATO:

150.0 is too high. The maximum allowed value is 101.0

250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

– Hans

checkmk-en mailing list

checkmk-en@lists.mathias-kettner.de

http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en

3

Hi Marcel,

Thanks. I was afraid of that.
I find that some processes (virtual machines and java come to mind) claim
(virtual) memory in such quantities that do not work well with this logic.
101.0 it is then. A pity.

-- Hans

···

Marcel Schulte wrote on 20160330:

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will
never get >100%. 101.0 is used to disable the threshold. If you want to use
the "old" thresholds in the new linux memory check you've to set param
"Upper levels for Total Data in relation to RAM" I think.

HTH,
Marcel

Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. M�rz 2016 um > 13:13 Uhr:

> Hi CMK list,
>
> I'm trying to tune rules for Virtual Memory Warning and Alerting level.
> In 1.2.8b7 these default to 80% and 90%
>
> I want to set them to 150% and 250% (I think this makes sense on KVM
> hypervisors)
>
> However I get these error in WATO:
> 150.0 is too high. The maximum allowed value is 101.0
> 250.0 is too high. The maximum allowed value is 101.0
>
> wtf ?! Can I be in control of levels myself ?
>
> Has anyone else seen this ? How to fix ?
>
> -- Hans

4

Hi Hans,

I don’t understand your issue.

Assumed a server with 8GB RAM and 16GB SWAP, you can set any of these params (or any combination of them):

  • set param “Level for RAM” to 80.0%/90.0% to get WARN/CRIT at 6.4GB/7.2GB of physical RAM used

  • set param “Level for Swap” to 10.0%/20.0% to get WARN/CRIT at 1.6GB/3.2GB of swap space used

  • set param “Levels for Total virtual memory” to 80.0%/90.0% to get WARN/CRIT at 19.2GB/21.6GB of all available memory (RAM+SWAP) used

  • set param “Upper levels for Total Data in relation to RAM” to 150.0%/250.0% to get WARN/CRIT at 12.0GB/20.0GB of all available memory used

What’s the problem with that?

And how would it be possible to use >100% of all available virtual memory (remember: virtual memory = RAM + SWAP)?

Regards,

Marcel

···

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um 14:02 Uhr:

Hi Marcel,

Thanks. I was afraid of that.

I find that some processes (virtual machines and java come to mind) claim

(virtual) memory in such quantities that do not work well with this logic.

101.0 it is then. A pity.

– Hans

Marcel Schulte wrote on 20160330:

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will

never get >100%. 101.0 is used to disable the threshold. If you want to use

the “old” thresholds in the new linux memory check you’ve to set param

“Upper levels for Total Data in relation to RAM” I think.

HTH,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um > > > 13:13 Uhr:

Hi CMK list,

I’m trying to tune rules for Virtual Memory Warning and Alerting level.

In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on KVM

hypervisors)

However I get these error in WATO:

150.0 is too high. The maximum allowed value is 101.0

250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

– Hans

5

Looking at the rules in 1.2.8b8.cre, it looks like there are more flexible
options for memory and swap monitoring than there are in 1.2.6p16.cre.

···

Sent from my iPad

On Mar 30, 2016, at 8:02 AM, Hans Lambermont <hans@shapeways.com> wrote:

Hi Marcel,

Thanks. I was afraid of that.
I find that some processes (virtual machines and java come to mind) claim
(virtual) memory in such quantities that do not work well with this logic.
101.0 it is then. A pity.

-- Hans

Marcel Schulte wrote on 20160330:

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will
never get >100%. 101.0 is used to disable the threshold. If you want to use
the "old" thresholds in the new linux memory check you've to set param
"Upper levels for Total Data in relation to RAM" I think.

HTH,
Marcel

Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. März 2016 um >> 13:13 Uhr:

Hi CMK list,

I'm trying to tune rules for Virtual Memory Warning and Alerting level.
In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on KVM
hypervisors)

However I get these error in WATO:
   150.0 is too high. The maximum allowed value is 101.0
   250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

-- Hans

_______________________________________________
checkmk-en mailing list
checkmk-en@lists.mathias-kettner.de
http://lists.mathias-kettner.de/mailman/listinfo/checkmk-en

6

Hi Marcel,

I'll given an example;

On a KVM machine with 128 GiB ram and (only) 8 GiB swap I run a few virtual
machines that together claim about 160 GiB ram ('VIRT' in top, 'VSZ' in ps),
yet if I add up the committed memory ('RES' in top, 'RSS' in ps, 'Memory
Committing' graph in 1.2.8b7) then they together use only 70 GiB.
This is all fine for this machine IMO.

VSZ:
ps aux | awk '{ s += $5 } END {print "sum =", s, s/1024^2, "GiB, average =", s/NR, ", over", NR, "counts" }'
sum = 164325620 156.713 GiB, average = 193780 , over 848 counts

RSS:
ps aux | awk '{ s += $6 } END {print "sum =", s, s/1024^2, "GiB, average =", s/NR, ", over", NR, "counts" }'
sum = 69238760 66.0312 GiB, average = 81745.9 , over 847 counts

free -m:
             total used free shared buffers cached
Mem: 128929 106939 21990 1 103 682
-/+ buffers/cache: 106153 22775
Swap: 8191 229 7962

So my VSZ is quite above the machines' RAM + SWAP. A factor 1.24 . That's why I
want my warning level to be at about 150% . I hope this helps.

-- Hans

···

Marcel Schulte wrote on 20160330:

Hi Hans,

I don't understand your issue.

Assumed a server with 8GB RAM and 16GB SWAP, you can set any of these
params (or any combination of them):
* set param "Level for RAM" to 80.0%/90.0% to get WARN/CRIT at 6.4GB/7.2GB
of physical RAM used
* set param "Level for Swap" to 10.0%/20.0% to get WARN/CRIT at 1.6GB/3.2GB
of swap space used
* set param "Levels for Total virtual memory" to 80.0%/90.0% to get
WARN/CRIT at 19.2GB/21.6GB of all available memory (RAM+SWAP) used
* set param "Upper levels for Total Data in relation to RAM" to
150.0%/250.0% to get WARN/CRIT at 12.0GB/20.0GB of all available memory used

What's the problem with that?

And how would it be possible to use >100% of all available virtual memory
(remember: virtual memory = RAM + SWAP)?

Regards,
Marcel

Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. M�rz 2016 um > 14:02 Uhr:

> Hi Marcel,
>
> Thanks. I was afraid of that.
> I find that some processes (virtual machines and java come to mind) claim
> (virtual) memory in such quantities that do not work well with this logic.
> 101.0 it is then. A pity.
>
> -- Hans
>
> Marcel Schulte wrote on 20160330:
> > AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will
> > never get >100%. 101.0 is used to disable the threshold. If you want to
> use
> > the "old" thresholds in the new linux memory check you've to set param
> > "Upper levels for Total Data in relation to RAM" I think.
> >
> > HTH,
> > Marcel
> >
> > Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. M�rz 2016 um > > > 13:13 Uhr:
> >
> > > Hi CMK list,
> > >
> > > I'm trying to tune rules for Virtual Memory Warning and Alerting level.
> > > In 1.2.8b7 these default to 80% and 90%
> > >
> > > I want to set them to 150% and 250% (I think this makes sense on KVM
> > > hypervisors)
> > >
> > > However I get these error in WATO:
> > > 150.0 is too high. The maximum allowed value is 101.0
> > > 250.0 is too high. The maximum allowed value is 101.0
> > >
> > > wtf ?! Can I be in control of levels myself ?
> > >
> > > Has anyone else seen this ? How to fix ?
> > >
> > > -- Hans
>

--
Hans Lambermont | Senior Architect
(t) +31407370104 (w) www.shapeways.com

7

Hi Hans,

  1. Memory info is grepped from /proc/meminfo - the information provided by the OS.

  2. When you have a look at “~/share/check_mk/checks/mem” you see this:

Total virtual memory

meminfo[“TotalTotal”] = meminfo[“MemTotal”] + meminfo[“SwapTotal”]

meminfo[“TotalUsed”] = meminfo[“MemUsed”] + meminfo[“SwapUsed”]

Could you please check your /proc/meminfo for the VSZ info?

VSZ is just the maximum size a process can get, initially set. It does not mean the process will get that size - the kernel schedules as much free mem as possible, if requested.

Regards,

Marcel

···

Hans Lambermont hans@shapeways.com schrieb am Fr., 1. Apr. 2016 um 14:44 Uhr:

Hi Marcel,

I’ll given an example;

On a KVM machine with 128 GiB ram and (only) 8 GiB swap I run a few virtual

machines that together claim about 160 GiB ram (‘VIRT’ in top, ‘VSZ’ in ps),

yet if I add up the committed memory (‘RES’ in top, ‘RSS’ in ps, 'Memory

Committing’ graph in 1.2.8b7) then they together use only 70 GiB.

This is all fine for this machine IMO.

VSZ:

ps aux | awk ‘{ s += $5 } END {print “sum =”, s, s/1024^2, “GiB, average =”, s/NR, “, over”, NR, “counts” }’

sum = 164325620 156.713 GiB, average = 193780 , over 848 counts

RSS:

ps aux | awk ‘{ s += $6 } END {print “sum =”, s, s/1024^2, “GiB, average =”, s/NR, “, over”, NR, “counts” }’

sum = 69238760 66.0312 GiB, average = 81745.9 , over 847 counts

free -m:

         total       used       free     shared    buffers     cached

Mem: 128929 106939 21990 1 103 682

-/+ buffers/cache: 106153 22775

Swap: 8191 229 7962

So my VSZ is quite above the machines’ RAM + SWAP. A factor 1.24 . That’s why I

want my warning level to be at about 150% . I hope this helps.

– Hans

Marcel Schulte wrote on 20160330:

Hi Hans,

I don’t understand your issue.

Assumed a server with 8GB RAM and 16GB SWAP, you can set any of these

params (or any combination of them):

  • set param “Level for RAM” to 80.0%/90.0% to get WARN/CRIT at 6.4GB/7.2GB

of physical RAM used

  • set param “Level for Swap” to 10.0%/20.0% to get WARN/CRIT at 1.6GB/3.2GB

of swap space used

  • set param “Levels for Total virtual memory” to 80.0%/90.0% to get

WARN/CRIT at 19.2GB/21.6GB of all available memory (RAM+SWAP) used

  • set param “Upper levels for Total Data in relation to RAM” to

150.0%/250.0% to get WARN/CRIT at 12.0GB/20.0GB of all available memory used

What’s the problem with that?

And how would it be possible to use >100% of all available virtual memory

(remember: virtual memory = RAM + SWAP)?

Regards,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um > > > 14:02 Uhr:

Hi Marcel,

Thanks. I was afraid of that.

I find that some processes (virtual machines and java come to mind) claim

(virtual) memory in such quantities that do not work well with this logic.

101.0 it is then. A pity.

– Hans

Marcel Schulte wrote on 20160330:

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage will

never get >100%. 101.0 is used to disable the threshold. If you want to

use

the “old” thresholds in the new linux memory check you’ve to set param

“Upper levels for Total Data in relation to RAM” I think.

HTH,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um > > > > > 13:13 Uhr:

Hi CMK list,

I’m trying to tune rules for Virtual Memory Warning and Alerting level.

In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on KVM

hypervisors)

However I get these error in WATO:

150.0 is too high. The maximum allowed value is 101.0

250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

– Hans

–

Hans Lambermont | Senior Architect

(t) +31407370104 (w) www.shapeways.com

8

Hi Marcel,

Sorry for late reply

Here's /proc/meminfo :

MemTotal: 132023468 kB
MemFree: 5333280 kB
Buffers: 41830240 kB
Cached: 773848 kB
SwapCached: 41004 kB
Active: 51165324 kB
Inactive: 50412248 kB
Active(anon): 50396468 kB
Inactive(anon): 8588524 kB
Active(file): 768856 kB
Inactive(file): 41823724 kB
Unevictable: 21068 kB
Mlocked: 21068 kB
SwapTotal: 8388604 kB
SwapFree: 8156164 kB
Dirty: 52 kB
Writeback: 0 kB
AnonPages: 58984176 kB
Mapped: 22732 kB
Shmem: 2772 kB
Slab: 22968020 kB
SReclaimable: 1176196 kB
SUnreclaim: 21791824 kB
KernelStack: 9136 kB
PageTables: 167864 kB
NFS_Unstable: 0 kB
Bounce: 0 kB
WritebackTmp: 0 kB
CommitLimit: 74400336 kB
Committed_AS: 82352332 kB
VmallocTotal: 34359738367 kB
VmallocUsed: 1323284 kB
VmallocChunk: 34354838132 kB
HardwareCorrupted: 0 kB
AnonHugePages: 6533120 kB
HugePages_Total: 0
HugePages_Free: 0
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB
DirectMap4k: 189924 kB
DirectMap2M: 134014976 kB

-- Hans

···

Marcel Schulte wrote on 20160401:

Hi Hans,

1. Memory info is grepped from /proc/meminfo - the information provided by
the OS.
2. When you have a look at "~/share/check_mk/checks/mem" you see this:
# Total virtual memory
meminfo["TotalTotal"] = meminfo["MemTotal"] + meminfo["SwapTotal"]
meminfo["TotalUsed"] = meminfo["MemUsed"] + meminfo["SwapUsed"]

Could you please check your /proc/meminfo for the VSZ info?

VSZ is just the maximum size a process can get, initially set. It does not
mean the process will get that size - the kernel schedules as much free mem
as possible, if requested.

Regards,
Marcel

Hans Lambermont <hans@shapeways.com> schrieb am Fr., 1. Apr. 2016 um > 14:44 Uhr:

> Hi Marcel,
>
> I'll given an example;
>
> On a KVM machine with 128 GiB ram and (only) 8 GiB swap I run a few virtual
> machines that together claim about 160 GiB ram ('VIRT' in top, 'VSZ' in
> ps),
> yet if I add up the committed memory ('RES' in top, 'RSS' in ps, 'Memory
> Committing' graph in 1.2.8b7) then they together use only 70 GiB.
> This is all fine for this machine IMO.
>
> VSZ:
> ps aux | awk '{ s += $5 } END {print "sum =", s, s/1024^2, "GiB, average
> =", s/NR, ", over", NR, "counts" }'
> sum = 164325620 156.713 GiB, average = 193780 , over 848 counts
>
> RSS:
> ps aux | awk '{ s += $6 } END {print "sum =", s, s/1024^2, "GiB, average
> =", s/NR, ", over", NR, "counts" }'
> sum = 69238760 66.0312 GiB, average = 81745.9 , over 847 counts
>
> free -m:
> total used free shared buffers cached
> Mem: 128929 106939 21990 1 103 682
> -/+ buffers/cache: 106153 22775
> Swap: 8191 229 7962
>
> So my VSZ is quite above the machines' RAM + SWAP. A factor 1.24 . That's
> why I
> want my warning level to be at about 150% . I hope this helps.
>
> -- Hans
>
> Marcel Schulte wrote on 20160330:
>
> > Hi Hans,
> >
> > I don't understand your issue.
> >
> > Assumed a server with 8GB RAM and 16GB SWAP, you can set any of these
> > params (or any combination of them):
> > * set param "Level for RAM" to 80.0%/90.0% to get WARN/CRIT at
> 6.4GB/7.2GB
> > of physical RAM used
> > * set param "Level for Swap" to 10.0%/20.0% to get WARN/CRIT at
> 1.6GB/3.2GB
> > of swap space used
> > * set param "Levels for Total virtual memory" to 80.0%/90.0% to get
> > WARN/CRIT at 19.2GB/21.6GB of all available memory (RAM+SWAP) used
> > * set param "Upper levels for Total Data in relation to RAM" to
> > 150.0%/250.0% to get WARN/CRIT at 12.0GB/20.0GB of all available memory
> used
> >
> > What's the problem with that?
> >
> > And how would it be possible to use >100% of all available virtual memory
> > (remember: virtual memory = RAM + SWAP)?
> >
> > Regards,
> > Marcel
> >
> > Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. M�rz 2016 um > > > 14:02 Uhr:
> >
> > > Hi Marcel,
> > >
> > > Thanks. I was afraid of that.
> > > I find that some processes (virtual machines and java come to mind)
> claim
> > > (virtual) memory in such quantities that do not work well with this
> logic.
> > > 101.0 it is then. A pity.
> > >
> > > -- Hans
> > >
> > > Marcel Schulte wrote on 20160330:
> > > > AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage
> will
> > > > never get >100%. 101.0 is used to disable the threshold. If you want
> to
> > > use
> > > > the "old" thresholds in the new linux memory check you've to set
> param
> > > > "Upper levels for Total Data in relation to RAM" I think.
> > > >
> > > > HTH,
> > > > Marcel
> > > >
> > > > Hans Lambermont <hans@shapeways.com> schrieb am Mi., 30. M�rz 2016 > > um > > > > > 13:13 Uhr:
> > > >
> > > > > Hi CMK list,
> > > > >
> > > > > I'm trying to tune rules for Virtual Memory Warning and Alerting
> level.
> > > > > In 1.2.8b7 these default to 80% and 90%
> > > > >
> > > > > I want to set them to 150% and 250% (I think this makes sense on
> KVM
> > > > > hypervisors)
> > > > >
> > > > > However I get these error in WATO:
> > > > > 150.0 is too high. The maximum allowed value is 101.0
> > > > > 250.0 is too high. The maximum allowed value is 101.0
> > > > >
> > > > > wtf ?! Can I be in control of levels myself ?
> > > > >
> > > > > Has anyone else seen this ? How to fix ?
> > > > >
> > > > > -- Hans
> > >
>
> --
> Hans Lambermont | Senior Architect
> (t) +31407370104 (w) www.shapeways.com
>

--
Hans Lambermont | Senior Architect
(t) +31407370104 (w) www.shapeways.com

9

Hi Hans,

in you cannot tell me which values represent your needs I cannot help.

Regards,
Marcel

···

Hans Lambermont hans@shapeways.com schrieb am Mi., 6. Apr. 2016 17:22:

Hi Marcel,

Sorry for late reply

Here’s /proc/meminfo :

MemTotal: 132023468 kB

MemFree: 5333280 kB

Buffers: 41830240 kB

Cached: 773848 kB

SwapCached: 41004 kB

Active: 51165324 kB

Inactive: 50412248 kB

Active(anon): 50396468 kB

Inactive(anon): 8588524 kB

Active(file): 768856 kB

Inactive(file): 41823724 kB

Unevictable: 21068 kB

Mlocked: 21068 kB

SwapTotal: 8388604 kB

SwapFree: 8156164 kB

Dirty: 52 kB

Writeback: 0 kB

AnonPages: 58984176 kB

Mapped: 22732 kB

Shmem: 2772 kB

Slab: 22968020 kB

SReclaimable: 1176196 kB

SUnreclaim: 21791824 kB

KernelStack: 9136 kB

PageTables: 167864 kB

NFS_Unstable: 0 kB

Bounce: 0 kB

WritebackTmp: 0 kB

CommitLimit: 74400336 kB

Committed_AS: 82352332 kB

VmallocTotal: 34359738367 kB

VmallocUsed: 1323284 kB

VmallocChunk: 34354838132 kB

HardwareCorrupted: 0 kB

AnonHugePages: 6533120 kB

HugePages_Total: 0

HugePages_Free: 0

HugePages_Rsvd: 0

HugePages_Surp: 0

Hugepagesize: 2048 kB

DirectMap4k: 189924 kB

DirectMap2M: 134014976 kB

– Hans

Marcel Schulte wrote on 20160401:

Hi Hans,

  1. Memory info is grepped from /proc/meminfo - the information provided by

the OS.

  1. When you have a look at “~/share/check_mk/checks/mem” you see this:

Total virtual memory

meminfo[“TotalTotal”] = meminfo[“MemTotal”] + meminfo[“SwapTotal”]

meminfo[“TotalUsed”] = meminfo[“MemUsed”] + meminfo[“SwapUsed”]

Could you please check your /proc/meminfo for the VSZ info?

VSZ is just the maximum size a process can get, initially set. It does not

mean the process will get that size - the kernel schedules as much free mem

as possible, if requested.

Regards,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Fr., 1. Apr. 2016 um > > > 14:44 Uhr:

Hi Marcel,

I’ll given an example;

On a KVM machine with 128 GiB ram and (only) 8 GiB swap I run a few virtual

machines that together claim about 160 GiB ram (‘VIRT’ in top, ‘VSZ’ in

ps),

yet if I add up the committed memory (‘RES’ in top, ‘RSS’ in ps, 'Memory

Committing’ graph in 1.2.8b7) then they together use only 70 GiB.

This is all fine for this machine IMO.

VSZ:

ps aux | awk '{ s += $5 } END {print “sum =”, s, s/1024^2, "GiB, average

=", s/NR, “, over”, NR, “counts” }’

sum = 164325620 156.713 GiB, average = 193780 , over 848 counts

RSS:

ps aux | awk '{ s += $6 } END {print “sum =”, s, s/1024^2, "GiB, average

=", s/NR, “, over”, NR, “counts” }’

sum = 69238760 66.0312 GiB, average = 81745.9 , over 847 counts

free -m:

         total       used       free     shared    buffers     cached

Mem: 128929 106939 21990 1 103 682

-/+ buffers/cache: 106153 22775

Swap: 8191 229 7962

So my VSZ is quite above the machines’ RAM + SWAP. A factor 1.24 . That’s

why I

want my warning level to be at about 150% . I hope this helps.

– Hans

Marcel Schulte wrote on 20160330:

Hi Hans,

I don’t understand your issue.

Assumed a server with 8GB RAM and 16GB SWAP, you can set any of these

params (or any combination of them):

  • set param “Level for RAM” to 80.0%/90.0% to get WARN/CRIT at

6.4GB/7.2GB

of physical RAM used

  • set param “Level for Swap” to 10.0%/20.0% to get WARN/CRIT at

1.6GB/3.2GB

of swap space used

  • set param “Levels for Total virtual memory” to 80.0%/90.0% to get

WARN/CRIT at 19.2GB/21.6GB of all available memory (RAM+SWAP) used

  • set param “Upper levels for Total Data in relation to RAM” to

150.0%/250.0% to get WARN/CRIT at 12.0GB/20.0GB of all available memory

used

What’s the problem with that?

And how would it be possible to use >100% of all available virtual memory

(remember: virtual memory = RAM + SWAP)?

Regards,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 um > > > > > 14:02 Uhr:

Hi Marcel,

Thanks. I was afraid of that.

I find that some processes (virtual machines and java come to mind)

claim

(virtual) memory in such quantities that do not work well with this

logic.

101.0 it is then. A pity.

– Hans

Marcel Schulte wrote on 20160330:

AFAIK virtual memory is defined as RAM+SWAP, so the percentual usage

will

never get >100%. 101.0 is used to disable the threshold. If you want

to

use

the “old” thresholds in the new linux memory check you’ve to set

param

“Upper levels for Total Data in relation to RAM” I think.

HTH,

Marcel

Hans Lambermont hans@shapeways.com schrieb am Mi., 30. März 2016 > > > > um > > > > > > > 13:13 Uhr:

Hi CMK list,

I’m trying to tune rules for Virtual Memory Warning and Alerting

level.

In 1.2.8b7 these default to 80% and 90%

I want to set them to 150% and 250% (I think this makes sense on

KVM

hypervisors)

However I get these error in WATO:

150.0 is too high. The maximum allowed value is 101.0

250.0 is too high. The maximum allowed value is 101.0

wtf ?! Can I be in control of levels myself ?

Has anyone else seen this ? How to fix ?

– Hans

–

Hans Lambermont | Senior Architect

(t) +31407370104 (w) www.shapeways.com

–

Hans Lambermont | Senior Architect

(t) +31407370104 (w) www.shapeways.com