Weird issue with hostmib

[mrossTTX]

ok im running cacti 1.2.22 on a windows server. these graphs are looking at the server cacti is running on.
the upper graph is using the " Host MIB- Processes", where it normally looks at the PID, but I changed it so it would look at the process NAME.. spoolsv.exe and determine how much memory its using. I have a thold that says if its less than 1, it's a breach and notify me, then run a command to start that service. This works perfectly.

However, I think the other host mib metric would be better to use. this is the lower graph. its using the Host MIB - Total App Running Metric. (I also used the Host MIB Total App Memory metric and had the same results that are troubling).

So, the issue I am seeing is that using the Total App Running metric, doesn't always pick up when the service, or process is running.. the only one that's reliable is the Host MIB processes metric. the others seem to just stop graphing as some point after i stop the service i am testing with.. even if i stop and disable this service. sometimes the graph just keeps graphing data even though the service is not running.
Likewise when the service is running, or restarted, the Total App Running or Total App Memory metrics dont pick up any data and thus dont graph anything..
but wait there's more..

without touching anything in Cacti, it would suddenly start graphing data again, while the top graph never stopped graphing.
The blank areas of the top graph are times when i stopped the service to demonstrate it would detect it.

Any ideas on where to start checking in my setup.. something i may have misconfigured maybe?

[mrossTTX]

case in point..here are the thold times watching these.. note the top 2 are watching the process NAME.. the graph that works fine.
the last one shows the total app running metric.. notice the whole time it says it was not running.. zero data the other graphs show data.. then randomly without me touching cacti or the server, it starts pulling data?

[TheWitness]

Well, the if the service is down, I mean hmib, the entries don't disappear from the tables, they stay at the last stats.

As far as things not updating, you need to look at the collection frequencies in on the Settings > HMIB tab. I can not judge you setup. You might want to increase your SNMP timeouts a bit as that's super important and Windows his horrible on timeouts especially to disk OID's. Upgrade to 1.2.23 if you can manage it.

[mrossTTX]

took your advise to heart. made changes to the hostmib collection frequencies, as shown below.
Also for this one host,which is the localhost that cacti is running on, increased its SNMP timeouts, as shown blelow.
Notice, the two hostmib graphs dont update the same during the time period that the service was stopped (i disabled it for this time period to demonstrate, then enabled and started it to show the difference between these two).
I would think that since both are using hostmib templates and graphs they would update the same way.

[mrossTTX]

by way of comparison for when the service is stopped.. i stopped and disabled it, the process host mib graph recognizes the process is no longer running, while the host mib total running for that same service name, has yet to see it's not running. as of posting this, its been over 15 minutes and the total running host mib has not seen that it's stopped yet

[mrossTTX]

I also think part of the confusion (for me) here is the graph that updates correctly is using SNMP data, the other is using the script server as shown below (screen shot of the data sources)

[mrossTTX]

this time it took 25 minutes for the host mib total running to see the service was stopped.

[TheWitness]

You should track the data collector, and see that it's properly running every minute. What is the Cacti Data collector frequency? What do the HMIB STATS: look like in the Cacti.log?

[TheWitness]

Grab a fresh copy of the hmib plugin and then put the Cacti Data Collector in Medium level. You should then see the times when the collection is running.

[mrossTTX]

git hub shows v 3.3 and my cacti shows v 3.3 for hmib.
are the files on github new, but the version not updated?
also here is my current log file for hmib...

[mrossTTX]

ok i redid the hmib plug in by downloading it again and removing\reinstalling.
created a graph for the spooler service that im testing my monitoring with.
as you can see, the service is stopped and disabled, but hmib still thinks its running

[mrossTTX]

i put logging to medium and told it to look at the hmib plug in.. below is what the log starts to show me immediately.

[mrossTTX]

got upgraded to 1.2.23 today.
i noticed HMIB still is not acting properly.
in fact, now when i enable logging to medium or anything higher than Low, i see this every few seconds in my log..
2023-01-10 12:39:30 - PLUGIN WARNING: Function does not exist config_form with function cycle_config_form
2023-01-10 12:39:29 - PLUGIN WARNING: Function does not exist config_form with function cycle_config_form
2023-01-10 12:39:29 - PLUGIN WARNING: Function does not exist config_form with function cycle_config_form
2023-01-10 12:39:29 - PLUGIN WARNING: Function does not exist config_form with function cycle_config_form
2023-01-10 12:39:28 - PLUGIN WARNING: Function does not exist config_form with function cycle_config_form

now.. i did have cycle installed, but hadnt started using it yet.. so i disabled the cycle plug in (not that i think it has anything to do with HMIB directly.. but the errors continue, and the graph i am testing with continues to not update in real time.

[mrossTTX]

i looked in the log, and i set it for Errors++
this is what i see in there.

[mrossTTX]

switched to look at SQL (i am just going through each log type... and i see)