Dear friends,
For a while I tried to debug this for myself - so far unsuccessfully.
1. System:
- Zabbix v4.4.7 (I know it is not officially supported any longer) installed on Ubuntu 18.04 LTS Server
- Agent installed on Fujitsu PRIMERGY RX2530 M4, running Windows Server 2016, containing a RAID controller allowing for max 8 disks, but only equipped with 4 disks (2x RAID-1) => C:/ and D:/
- Template Module Windows physical disks by Zabbix agent: Physical disks WMI get: Physical disks discovery
2. What it does correclty:
2.a. On Windows Server 2016 => Powershell:
=> this returns all the expected disks, namely:
- Name: "0 C:" (along with all the expected attibutes, such as PrecentIdleTime, etc.)
- Name: "1 D:"
- I have not been able to observe the other disks mentioned below (yet), but I am confident these would also appear if only I was lucky enough to sample at the right time
2.b. Latest data => {Hostname}: Physical disks WMI get
=> here I receive mostly the same content as mentioned above (i.e. "0 C:" and "1 D:"), expect...
3. What is unexpected:
along with 2.b., I sometimes also observer the following Names in addition:
- "Name", "0 C:" => 50 out of 50 times
- "Name", "1 D:" => 50 out of 50 times
- "Name", "2" => 0...5 out of 50 times (expected: 0 times)
- "Name", "3" => 0...3 out of 50 times (expected: 0 times)
- "Name", "4" => 0...2 out of 50 times (expected: 0 times)
- "Name", "5" => 0...1 out of 50 times (expected: 0 times)
In the Latest data, I notice some big gaps for drives 2...5 ranging from a few minutes to > 12 hours.
However, there is a tendency for the drives being discovered around the full hour.
4. How this affects me:
Since the not mounted disks (2...5) only appear sporadically, the last value for the (automatically generated, via discovery rule) item "Disk utilization" might happen to be above the threshold for the trigger "Disk is overloaded" (i.e. 95% during 15m), before this drive will be discovered again. This frequently fires triggers.
5. Questions:
5.a. Does anybody have any idea what might this WMI call cause to return not mounted disks to appear?
5.b. How can I prevent these unmounted disks from causing issues?
5.c. What kind of further info would be helpful to identify root causes?
6. Going furher:
So far, I can think of the following possibilities:
6.a. disable the trigger (also for C: and D: - not ideal
6.b. applying a RegEx filter (Physical disks discovery => Filters => Does not match)
currently, this is set to "_Total".
=> Am I right to assume it would work with "_Total|2|3|4|5"?
6.c. looking for a different WMI call and implementing it instead (I had to do someting similar for Network Interfaces discovery as a workaround, see here)
For a while I tried to debug this for myself - so far unsuccessfully.
1. System:
- Zabbix v4.4.7 (I know it is not officially supported any longer) installed on Ubuntu 18.04 LTS Server
- Agent installed on Fujitsu PRIMERGY RX2530 M4, running Windows Server 2016, containing a RAID controller allowing for max 8 disks, but only equipped with 4 disks (2x RAID-1) => C:/ and D:/
- Template Module Windows physical disks by Zabbix agent: Physical disks WMI get: Physical disks discovery
2. What it does correclty:
2.a. On Windows Server 2016 => Powershell:
Code:
Get-WmiObject win32_perfformatteddata_perdisk_physicaldisk
- Name: "0 C:" (along with all the expected attibutes, such as PrecentIdleTime, etc.)
- Name: "1 D:"
- I have not been able to observe the other disks mentioned below (yet), but I am confident these would also appear if only I was lucky enough to sample at the right time
2.b. Latest data => {Hostname}: Physical disks WMI get
=> here I receive mostly the same content as mentioned above (i.e. "0 C:" and "1 D:"), expect...
3. What is unexpected:
along with 2.b., I sometimes also observer the following Names in addition:
- "Name", "0 C:" => 50 out of 50 times
- "Name", "1 D:" => 50 out of 50 times
- "Name", "2" => 0...5 out of 50 times (expected: 0 times)
- "Name", "3" => 0...3 out of 50 times (expected: 0 times)
- "Name", "4" => 0...2 out of 50 times (expected: 0 times)
- "Name", "5" => 0...1 out of 50 times (expected: 0 times)
In the Latest data, I notice some big gaps for drives 2...5 ranging from a few minutes to > 12 hours.
However, there is a tendency for the drives being discovered around the full hour.
4. How this affects me:
Since the not mounted disks (2...5) only appear sporadically, the last value for the (automatically generated, via discovery rule) item "Disk utilization" might happen to be above the threshold for the trigger "Disk is overloaded" (i.e. 95% during 15m), before this drive will be discovered again. This frequently fires triggers.
5. Questions:
5.a. Does anybody have any idea what might this WMI call cause to return not mounted disks to appear?
5.b. How can I prevent these unmounted disks from causing issues?
5.c. What kind of further info would be helpful to identify root causes?
6. Going furher:
So far, I can think of the following possibilities:
6.a. disable the trigger (also for C: and D: - not ideal
6.b. applying a RegEx filter (Physical disks discovery => Filters => Does not match)
currently, this is set to "_Total".
=> Am I right to assume it would work with "_Total|2|3|4|5"?
6.c. looking for a different WMI call and implementing it instead (I had to do someting similar for Network Interfaces discovery as a workaround, see here)
Comment