Question for check_linux_stats.pl

[michaelli]

Hi,

I configure check memory service in RHEL and oracle linux by check_linux_stats.pl. According to my observation, I find the memory used % is not accuracy. Below is the data that I capture from our servers. Kindly correct me if I mistake understand the concept.

1. Red Hat Linux
Kernel Vesion: Linux cmnndns01 2.6.32-431.17.1.el6.x86_64 #1 SMP Fri Apr 11 17:27:00 EDT 2014 x86_64 x86_64 x86_64 GNU/Linux

[root@cmnndns01 etc]# /usr/local/nagios/libexec/check_linux_stats.pl -M -w 90 -c 95
MEMORY OK : Mem used: 11.38%, Swap used: 0.00% |MemUsed=11.38%;90;95 SwapUsed=0.00%;; MemCached=10.78% SwapCached=0.00% Active=14.29%
From check_linux_stats.pl result, memory used is 11.38%

[root@cmnndns01 etc]# free -m
total used free shared buffers cached
Mem: 7870 1739 6130 0 409 848
-/+ buffers/cache: 482 7387
Swap: 8191 0 8191
Result from "free -m", 1739/7870 x 100% = 22%

07:00:01 AM kbmemfree kbmemused %memused kbbuffers kbcached kbcommit %commit
10:00:01 AM 6277964 1780976 22.10 419076 868072 480724 2.92
10:10:01 AM 6277872 1781068 22.10 419076 868080 482300 2.93
10:20:01 AM 6278252 1780688 22.10 419076 868100 431320 2.62
10:30:01 AM 6279136 1779804 22.08 419076 868104 430880 2.62
10:40:01 AM 6277896 1781044 22.10 419076 868116 481728 2.93
Result from sar -r, %memused = 22%

2. Oracle Linux
Linux lcsridb14 3.8.13-16.2.1.el6uek.x86_64 #1 SMP Thu Nov 7 17:01:44 PST 2013 x86_64 x86_64 x86_64 GNU/Linux

[root@lcsridb14 ~]# /usr/local/nagios/libexec/check_linux_stats.pl -M -w 90 -c 95
MEMORY OK : Mem used: 17.80%, Swap used: 0.00% |MemUsed=17.80%;90;95 SwapUsed=0.00%;; MemCached=81.36% SwapCached=0.00% Active=36.45%
From check_linux_stats.pl result, memory used is 17.80%

[root@lcsridb14 ~]# free -m
total used free shared buffers cached
Mem: 12015 11873 142 0 617 9776
-/+ buffers/cache: 1480 10535
Swap: 32767 0 32767
Result from "free -m", 11873/12015 x 100% = 98.8%

10:00:01 AM kbmemfree kbmemused %memused kbbuffers kbcached kbcommit %commit
10:05:01 AM 154448 12149708 98.74 629976 10008400 708344 1.54
10:10:01 AM 140128 12164028 98.86 630000 10009576 722492 1.58
10:15:01 AM 337788 11966368 97.25 630032 9789020 745932 1.63
10:20:01 AM 153108 12151048 98.76 630096 10010076 708320 1.54
10:25:01 AM 156284 12147872 98.73 630128 10005980 708308 1.54
10:30:01 AM 210552 12093604 98.29 630164 9888484 942148 2.05

Result from sar -r, %memused = 98.29%

If this script is not suitable to run for these two version of Linux because we are unable to get alert if they are in high memory usage.
Where can I download a correct scripts for check memory usage?

[ssax]

They are returning different values for me as well, the check_linux_stats.pl script is actually pulling from /proc/meminfo.

Can you validate the contents of meminfo with what you are seeing from free -m/sar -r?

Code: Select all

cat /proc/meminfo

[ssax]

I should note that it's actually the perl module Sys-Statistics-Linux-MemStats that is pulling from /proc/meminfo.

Code: Select all

$meminfo{memused}     = sprintf('%u', $meminfo{memtotal} - $meminfo{memfree});
$meminfo{memusedper}  = sprintf('%.2f', 100 * $meminfo{memused} / $meminfo{memtotal});

So it's doing the calculations directly from the /proc/meminfo output.

[michaelli]

Hi,

I think /proc/meminfo, sar -r and free -k are matched to each other, it should be calculation mistake in script

Code: Select all

[root@lcsridb14 libexec]# cat /proc/meminfo
MemTotal:       12304156 kB
MemFree:          128716 kB
Buffers:          639328 kB
Cached:          9960676 kB
SwapCached:           24 kB
Active:          4508564 kB
Inactive:        6717436 kB
Active(anon):     283776 kB
Inactive(anon):   342780 kB
Active(file):    4224788 kB
Inactive(file):  6374656 kB
Unevictable:           0 kB
Mlocked:               0 kB
SwapTotal:      33554428 kB
SwapFree:       33554352 kB
Dirty:             78052 kB
Writeback:             0 kB
AnonPages:        626388 kB
Mapped:            26828 kB
Shmem:               404 kB
Slab:             763844 kB
SReclaimable:     408412 kB
SUnreclaim:       355432 kB
KernelStack:        2648 kB
PageTables:         8196 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    39706504 kB
Committed_AS:     754468 kB
VmallocTotal:   34359738367 kB
VmallocUsed:      339000 kB
VmallocChunk:   34359325580 kB
HardwareCorrupted:     0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:       10588 kB
DirectMap2M:    12570624 kB

Code: Select all

10:30:01 AM kbmemfree kbmemused  %memused kbbuffers  kbcached  kbcommit   %commit
10:35:01 AM    152484  12151672     98.76    638996   9975480    712616      1.55
10:40:01 AM    157768  12146388     98.72    639036   9933704    751320      1.64
10:45:01 AM    154608  12149548     98.74    639056   9973228    712688      1.55
10:50:01 AM    156260  12147896     98.73    639072   9971924    712628      1.55
10:55:01 AM    131160  12172996     98.93    639084   9960088    751296      1.64
11:00:01 AM    155456  12148700     98.74    639112   9972548    712676      1.55
11:05:01 AM    156672  12147484     98.73    639168   9970224    712640      1.55
11:10:01 AM    157812  12146344     98.72    639204   9970288    712640      1.55
11:15:01 AM     96920  12207236     99.21    639228   9962144    946472      2.06
11:20:01 AM    147728  12156428     98.80    639248   9973440    722196      1.57
11:25:01 AM    153880  12150276     98.75    639256   9971168    717452      1.56
11:30:01 AM    149268  12154888     98.79    639292   9974720    718216      1.57
11:35:01 AM    357404  11946752     97.10    639328   9733316    754168      1.64
Average:       169680  12134476     98.62    637452   9950236    734261      1.60

Code: Select all

[root@lcsridb14 libexec]# free -k
             total       used       free     shared    buffers     cached
Mem:      12304156   11893816     410340        404     639340    9679356
-/+ buffers/cache:    1575120   10729036
Swap:     33554428         76   33554352

Code: Select all

[root@lcsridb14 libexec]# ./check_linux_stats.pl -M -w 90 -c 95
MEMORY OK : Mem used: 17.74%, Swap used: 0.00% |MemUsed=17.74%;90;95 SwapUsed=0.00%;; MemCached=81.05% SwapCached=0.00% Active=36.48%

[lmiltchev]

[root@cmnndns01 etc]# free -m
total used free shared buffers cached
Mem: 7870 1739 6130 0 409 848
-/+ buffers/cache: 482 7387
Swap: 8191 0 8191
Result from "free -m", 1739/7870 x 100% = 22%

The plugin subtracts the cached memory first, which explains the discrepancy.

( (free - cached) / total ) * 100%

( (1739 - 848) / 7870 ) * 100% = (891 / 7870) * 100% = 0.1132*100 = 11.3%

There is some rounding going on but that's pretty close.

[root@cmnndns01 etc]# /usr/local/nagios/libexec/check_linux_stats.pl -M -w 90 -c 95
MEMORY OK : Mem used: 11.38%, Swap used: 0.00% |MemUsed=11.38%;90;95 SwapUsed=0.00%;; MemCached=10.78% SwapCached=0.00% Active=14.29%
From check_linux_stats.pl result, memory used is 11.38%

[michaelli]

Hi lmiltchev,
Thank you for your explanation.

[tmcdonald]

Are we all clear to close this?