Ad Widget

Collapse

wrong disk size

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • pr0d1gy
    Member
    • Jun 2016
    • 31

    #1

    wrong disk size

    Hi ,
    I am new with Zabbix so I am sure I have a way to fix it but I couldn't find a solution yet .
    I receiving data from a Centos6.6 server on Zabbix 3.0.1 "Template SNMP OS Linux " .
    the problem is the data disk size is completely wrong .
    Code:
    	Allocation units for storage /stornext/Vault	4 KB
    	
    		Total disk space on /stornext/Vault 1.48 TB
    		Total disk space on /stornext/Vault in units 398.24 Munits
    		Used disk space on /stornext/Vault 4.21 TB
    		Used disk space on /stornext/Vault in units 1.13 Gunits
    while df -h /stornext/Vault/ on server shows
    Code:
    Filesystem           Size  Used Avail Use% Mounted on
    /dev/cvfsctl4_Vault   66T   37T   30T  56% /stornext/Vault
    what am I missing ? is there any parameter I need to change to get the right results ?

    Kind regards ,
    Pr0
  • glebs.ivanovskis
    Senior Member
    • Jul 2015
    • 237

    #2
    Seems like you are using 32 bit counters. Use 64 bit instead! With 4 KB block size 32 bit counter for blocks will be exhausted at 16 TB of storage space.

    Comment

    • pr0d1gy
      Member
      • Jun 2016
      • 31

      #3
      Originally posted by glebs.ivanovskis
      Seems like you are using 32 bit counters. Use 64 bit instead! With 4 KB block size 32 bit counter for blocks will be exhausted at 16 TB of storage space.
      Hi,

      thanks for your reply .
      I understand what are you saying but how do I implement it in zabbix?
      is it done by changing one of the parameters ?

      Regards ,
      Pr0

      Comment

      • pr0d1gy
        Member
        • Jun 2016
        • 31

        #4
        I really wonder ,
        I am the only one who bumped this issue ? no one seen it before ?

        Regard,
        pr0d

        Comment

        • kloczek
          Senior Member
          • Jun 2006
          • 1771

          #5
          Originally posted by pr0d1gy
          I really wonder ,
          I am the only one who bumped this issue ? no one seen it before ?
          Very possible because monitoring Linuxes over SNMP isn't best method of doing such things.
          Zabbix standard agent offers much better platform (uses less memory, consume much less CPU power and offers much better flexibility of monitoring anything without changing agent set up) .
          http://uk.linkedin.com/pub/tomasz-k%...zko/6/940/430/
          https://kloczek.wordpress.com/
          zapish - Zabbix API SHell binding https://github.com/kloczek/zapish
          My zabbix templates https://github.com/kloczek/zabbix-templates

          Comment

          • pr0d1gy
            Member
            • Jun 2016
            • 31

            #6
            Originally posted by kloczek
            Very possible because monitoring Linuxes over SNMP isn't best method of doing such things.
            Zabbix standard agent offers much better platform (uses less memory, consume much less CPU power and offers much better flexibility of monitoring anything without changing agent set up) .
            Thanks for your reply .
            I must say I have the same problem with zabbix agent over Mac .

            df -H on machine :
            Code:
            admins-Mac:~ admin$ df -H
            Filesystem                                  Size   Used  Avail Capacity iused      ifree %iused  Mounted on
            /dev/disk0s2                                 68G    13G    55G    20% 3275099   13418137   20%   /
            while zabbix monitor it :

            Code:
            	Filesystems (6 Items)
            	Free disk space on /
            vfs.fs.size[/,free]	60	7	365	Zabbix agent	2016-07-03 17:10:33	51.19 GB	-4 KB	Graph	
            	Free disk space on / (percentage)
            vfs.fs.size[/,pfree]	60	7	365	Zabbix agent	2016-07-03 17:10:34	80.69 %		Graph	
            	Free inodes on / (percentage)
            vfs.fs.inode[/,pfree]	60	7	365	Zabbix agent	2016-07-03 17:10:32	80.38 %		Graph	
            	Total Disk Size of /
            diskSize[/]	5	90		Zabbix agent	2016-07-03 17:10:54	0 2 68 13 55 20 3275105 13418131 20		History	
            	Total disk space on /
            vfs.fs.size[/,total]	3600	7	365	Zabbix agent	2016-07-03 17:10:00	63.68 GB		Graph	
            	Used disk space on /
            vfs.fs.size[/,used]	60	7	365	Zabbix agent	2016-07-03 17:10:36	12.25 GB	+4 KB	Graph
            as you can see 68GB of disk monitored as 63.68GB
            free space of 55GB monitored as 51.19GB
            consider monitoring TB sized and the offset will be huge .

            what can be wrong ?

            Comment

            • kloczek
              Senior Member
              • Jun 2006
              • 1771

              #7
              Originally posted by pr0d1gy
              as you can see 68GB of disk monitored as 63.68GB
              free space of 55GB monitored as 51.19GB
              consider monitoring TB sized and the offset will be huge .

              what can be wrong ?
              I have no idea.
              You should try to observe 'zabbix_agentd - "vfs.fs.size[/,free]"' running this command over truss to observe which one syscalls is using zabbix agent and what results are returning those syscalls on sampling disk space metrics.
              If you will find anything odd here just open a ticket on support.zabbix.com
              http://uk.linkedin.com/pub/tomasz-k%...zko/6/940/430/
              https://kloczek.wordpress.com/
              zapish - Zabbix API SHell binding https://github.com/kloczek/zapish
              My zabbix templates https://github.com/kloczek/zabbix-templates

              Comment

              • glebs.ivanovskis
                Senior Member
                • Jul 2015
                • 237

                #8
                Originally posted by pr0d1gy
                what can be wrong ?
                Nothing is wrong. df -H shows gigabytes, Zabbix shows gibibytes.

                Comment

                • pr0d1gy
                  Member
                  • Jun 2016
                  • 31

                  #9
                  Thanks for the heads up !

                  now it's more strange .
                  zabbix_get shows the right size
                  Code:
                  zabbix_get -s 172.16.50.82 -k vfs.fs.size[/,free]
                  54960041984
                  i't just the Latest Data monitor shows it wrong .
                  as far as I've tested seems zabbix_get get it's size like running df -H
                  Code:
                  df -H
                  Filesystem                                  Size   Used  Avail Capacity iused      ifree %iused  Mounted on
                  /dev/disk0s2                                 68G    13G    55G    20% 3279014   13414222   20%   /
                  while the latest monitor get it's result like running df -h
                  Code:
                  admins-Mac:~ admin$ df -h
                  Filesystem                                  Size   Used  Avail Capacity iused      ifree %iused  Mounted on
                  /dev/disk0s2                                64Gi   12Gi   51Gi    20% 3279012   13414224   20%   /
                  do you think I should open a ticket on it or there is some macro change I need to do?

                  Regards.

                  Comment

                  • pr0d1gy
                    Member
                    • Jun 2016
                    • 31

                    #10
                    Originally posted by glebs.ivanovskis
                    Nothing is wrong. df -H shows gigabytes, Zabbix shows gibibytes.
                    Ok
                    how can I make zabbix show size in gigabytes ?

                    Regards ,

                    Comment

                    • glebs.ivanovskis
                      Senior Member
                      • Jul 2015
                      • 237

                      #11
                      Originally posted by pr0d1gy
                      do you think I should open a ticket on it or there is some macro change I need to do?
                      There is no indication of a bug. Agent returns size in bytes, value is stored in DB in bytes, frontend uses 1024 as a kilo multiplier for bytes. The only thing is that frontend confuses users with K, M, G instead of Ki, Mi, Gi in this case and there are several related issues you can vote for:

                      And you can use a custom multiplier if you prefer GB to GiB.

                      Comment

                      • pr0d1gy
                        Member
                        • Jun 2016
                        • 31

                        #12
                        Thanks

                        can you point me how to custom multiplier it into GB ?
                        is there any manual for that?

                        Regards,

                        Comment

                        • glebs.ivanovskis
                          Senior Member
                          • Jul 2015
                          • 237

                          #13
                          Look here for "Use custom multiplier".

                          Comment

                          • pr0d1gy
                            Member
                            • Jun 2016
                            • 31

                            #14
                            Hi Glebs ,

                            I am almost there tho I get the same result .
                            I put "B" in Units and 1024 in Use custom multiplier
                            and still getting 63.68GB.
                            do you have a clue what is the right custom multiplier I need to write ?

                            Regards ,

                            Comment

                            • DmitryL
                              Senior Member
                              Zabbix Certified SpecialistZabbix Certified Professional
                              • May 2016
                              • 278

                              #15
                              He Pr0d1gy,

                              Please try this item configuration:


                              Best regards,
                              Dmitry

                              Comment

                              Working...