We have nearly 10 class C subnets (public IPs all) of which we use some (but never all) of the IPs for hosted servers.
We have neighboring subnets (I will omit some of the octets for obfuscation) which each have the same first 2 octets, and incremental 3rd octets before the 1-254 of the 4th octet.
x.x.178.1-254
x.x.179.1-254
x.x.180.1-254
x.x.181.1-254
x.x.182.1-254
x.x.183.1-254
Discovery is working grandly on ALL of these subnets except for the one running on 182.1-254. We have no idea why. Ping works to the IPs in that subnet. Hostname lookups work. zabbix_get even pulls the data we use for discovery. We have validated all of the steps we normally use against the other subnets and they work just fine. I've tried to recreate the rule, edit the timing, change things up. None of it works. It's all in vain. The only thing I haven't done is restart our Zabbix server.
What happens in the discovery process that I don't know about? I've tried zabbix_get. I presume it's similar if not identical code to how the discovery checks work if you're polling a zabbix_agent. Is there something I'm missing? Will zabbix mark an entire discovery rule as unusable if it hits a particular kind of IP or ping response with a certain number of IPs in that block? This block, unlike the others, has a certain few IPs which respond with "ICMP Host Unreachable from gateway x.x.x.x" though they are mixed with IPs which do respond properly. In spite of this, I tried to create a rule which had only 1 IP for the IP range. It was known good with ping and zabbix_get responding, but it never got discovered. Are there DNS oddities which can cause problems? We may have a few of those (acknowledged by our network team), but they might only affect a subset of these IPs.
I'm at my wit's end here. Has anyone else seen something like this? I really don't understand what's going on here.
We have neighboring subnets (I will omit some of the octets for obfuscation) which each have the same first 2 octets, and incremental 3rd octets before the 1-254 of the 4th octet.
x.x.178.1-254
x.x.179.1-254
x.x.180.1-254
x.x.181.1-254
x.x.182.1-254
x.x.183.1-254
Discovery is working grandly on ALL of these subnets except for the one running on 182.1-254. We have no idea why. Ping works to the IPs in that subnet. Hostname lookups work. zabbix_get even pulls the data we use for discovery. We have validated all of the steps we normally use against the other subnets and they work just fine. I've tried to recreate the rule, edit the timing, change things up. None of it works. It's all in vain. The only thing I haven't done is restart our Zabbix server.
What happens in the discovery process that I don't know about? I've tried zabbix_get. I presume it's similar if not identical code to how the discovery checks work if you're polling a zabbix_agent. Is there something I'm missing? Will zabbix mark an entire discovery rule as unusable if it hits a particular kind of IP or ping response with a certain number of IPs in that block? This block, unlike the others, has a certain few IPs which respond with "ICMP Host Unreachable from gateway x.x.x.x" though they are mixed with IPs which do respond properly. In spite of this, I tried to create a rule which had only 1 IP for the IP range. It was known good with ping and zabbix_get responding, but it never got discovered. Are there DNS oddities which can cause problems? We may have a few of those (acknowledged by our network team), but they might only affect a subset of these IPs.
I'm at my wit's end here. Has anyone else seen something like this? I really don't understand what's going on here.
Comment