zabbix rediscovers nodes from ip address that were deleted from discovery rule

[arapacana]

I have some ASR9ks that have many IP addresses configured. Zabbix's discovery process is pretty slow and does not seem to completely run through lists of IPs in a discovery rule -- it takes many runs (1 hr interval for discovery) over many days, and slowly new IPs for the same device show up. Zabbix will add the same host with _2 appended at the end, under the new IP. I will see this, and delete the IP from the discovery rule and delete the _2 host.

Normally this seems to mostly work... but often the same host will reappear with _2 appended, under an IP that was previously deleted from the discovery rule (and which is NOT in any other discovery rule). It does not appear to follow much of a pattern ... every 2-6 runs of the discovery rule interval it may find a few of the deleted IPs... but maybe not.

Here is one of the discovery rules that keeps re-adding hosts on deleted IPs. The IPs in this rule are:

172.20.91.2-254,
172.20.101.6-7,
172.20.101.11-13,
172.20.101.15-27,
172.20.101.30,
172.20.101.32-35,
172.20.101.37-39,
172.20.101.42,
172.20.101.44-47,
172.20.101.51-62,
172.20.101.64-69,
172.20.101.71-73,
172.20.101.75-79,
172.20.101.83-86,
172.20.101.88-91,
172.20.101.94-97,
172.20.101.99,
172.20.101.104,
172.20.101.106,
172.20.101.108-109,
172.20.101.111,
172.20.101.115-118,
172.20.101.120-121,
172.20.101.123-124,
172.20.101.126,
172.20.101.129-254

and the rest of the rule looks like:



Now in this case host X at 172.20.101.98, for example, keeps getting discovered and re-added with _2 appended to the end. The same happens with an identical rule that includes many public IPs, with the same uniqueness criteria and other settings as the rule noted here.

Two issues:
1. The uniqueness criteria of iso3.6.1.2.1.1.5.0 is the host's own hostname, so even if Zabbix found the host on a new IP in the rule it shouldn't add it as a separate host but add the IP as a secondary, which it does for some hosts, showing the IP indented below another in the Monitoring-->Discovery list.
2. The host simply shouldn't even be rediscovered on an IP that is not currently in the discovery rule.

I am at a loss as to how the deleted IPs are getting rediscovered since they are not in any discovery rule. Disabling the _2 hosts on their secondary IPs is not a good solution because the hosts still show up everywhere and the point is to keep things clean, where the host is only added under the main IP.

What am I missing, or is this somehow a bug?

Thanks for any help/insight,
-Seth

If it helps, the discovery action and operations (are as follows):

[Lucas09sm]

Zabbix periodically scans the IP ranges defined in 192.168.L.254 The frequency of the check is configurable for each rule individually.

Note that one discovery rule will always be processed by a single discoverer process. The IP range will not be split between multiple discoverer processes.

Each rule has a set of service checks defined to be performed for the IP range.

[arapacana]

Hmmm. Yes -- I did read the same documentation. Please note that the IPs in question are NOT a part of ANY discovery rule -- they have been deleted from a discovery rule and are still being discovered.

I have noticed that it seems like the earliest deleted and re-discovered IPs are no longer being re-discovered. It may be the case that only the more recently deleted IPs from the discovery rule are being re-discovered. I'll have to keep an eye on this to see if it's true.

I wish it was clear exactly when/how Zabbix ran through the actual discovery process, because literally every 1hr run of the rule it seems to find a few new IPs on the same hosts that it hadn't before. It's been over two weeks... why won't it correctly run through the entire IP ranges listed in one go, find ALL the IPs and perform the relevant defined operations in the discovery action? Why would it keep on newly finding IPs that have been in the range since the beginning? Why wouldn't it have run the discovery actions on the very first run?

Thanks,
-Seth