Synology Template Overhaul: SNMP issue tracking disks as LLD hosts
Hello,
I am currently trying to track all the disks of several Synology NAS systems in Zabbix.
The goal is to keep the NAS as the main host, while creating the individual disks as separate hosts through low-level discovery. This would allow us to track SMART values and attach disk-specific triggers and inventory information to each drive, and keep each information
The disk information is spread across several SNMP tables. Depending on the Synology model, I can receive:
Serial numbers
Model names
or
device names
SMART attribute names and raw values
The table positions cannot be used for matching, so the entries have to be correlated by their identity. For example:
This part is working reasonably well.
The problem is that the parent NAS host and the discovered disk hosts all use the same SNMPv3 interface and target IP. This unpleasantly leads to errors:
It appears that simply adjusting the check intervals is not a reliable solution. The conflict is related to the same SNMP source being monitored through multiple Zabbix hosts.
I mean, it is working ... somehow. The engineID will always collide somewhere else, so all metrics will finally arrive at some point. But I wouldn't call it "reliable"
Has anyone implemented a similar parent/child design, or can recommend the cleanest approach for this in Zabbix? I dont want to loose the individual tracking for the discs, because this way it is easier to view and monitor such assets.
I am currently trying to track all the disks of several Synology NAS systems in Zabbix.
The goal is to keep the NAS as the main host, while creating the individual disks as separate hosts through low-level discovery. This would allow us to track SMART values and attach disk-specific triggers and inventory information to each drive, and keep each information
The disk information is spread across several SNMP tables. Depending on the Synology model, I can receive:
Serial numbers
Model names
Code:
/dev/sdX
Code:
sataX
SMART attribute names and raw values
The table positions cannot be used for matching, so the entries have to be correlated by their identity. For example:
Code:
/dev/sda = sda = Drive 1
The problem is that the parent NAS host and the discovered disk hosts all use the same SNMPv3 interface and target IP. This unpleasantly leads to errors:
Code:
SNMP engineId is not unique
I mean, it is working ... somehow. The engineID will always collide somewhere else, so all metrics will finally arrive at some point. But I wouldn't call it "reliable"
Has anyone implemented a similar parent/child design, or can recommend the cleanest approach for this in Zabbix? I dont want to loose the individual tracking for the discs, because this way it is easier to view and monitor such assets.