Flowview dosen't kill flow process when disabled, or flow-capture.service is stoped

[arno-st]

Since I have some trouble with flowview at midnight, I decide to disable the plug in.
But still i get this error once in a while in cacti.og:
30/01/2025 07:16:07 - CMDPHP SQL Backtrace: (/plugins/flowview/flow_collector.php[812]:process_fv10(), /plugins/flowview/flow_collector.php[1685]:flowview_db_execute(), /plugins/flowview/database.php[70]:db_execute(), /lib/database.php[420]:db_execute_prepared()) 30/01/2025 07:16:07 - CMDPHP ERROR: A DB Exec Failed!, Error: You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near ' 0, 0, 'spr-itunes', '', 166, '2025-01-30 07:16:06.0', '2025-01-30 07:16:06.0...' at line 1 30/01/2025 07:16:07 - CMDPHP PHP ERROR WARNING Backtrace: (/plugins/flowview/flow_collector.php[812]:process_fv10(), /plugins/flowview/flow_collector.php[1605]:unpack(), CactiErrorHandler()) 30/01/2025 07:16:07 - ERROR PHP WARNING in Plugin 'flowview': unpack(): Type N: not enough input, need 4, have 1 in file: /usr/share/cacti/plugins/flowview/flow_collector.php on line: 1605
That only do half of the work, So I try to stop flow capture, but even after I did a 'systemctl stop flow-capture.service', all my flow_collector are still running and I have to kill them manualy.

What can I do ?

[TheWitness]

Post screen shots of your templates. It's likely that we are misprocessing one of them.

[arno-st]

Here it is:

[TheWitness]

Well, I'm thinking that there must be some code path that is not taking when the plugin itself is disabled. Right now, disabling stops hooks from firing, but does not stop the flow-capture service.

The thing at midnight is likely the IRR download. We should allow that to be disabled separately from the rest of the plugin. How big is your Cacti database and server? Cores + memory?

[arno-st]

The database servers, 2 system with MariaDB 10.6.17, and Galera
CPU 2x10 core @ 2.2 GHz
RAM 256 GiB
system Disk : 1*440TB ( os )
DATA Disk: 4 drive mirrored 2x2.91 TiB, 5.9TB usable.

The flowview DB is 18Gig
The cacti DB is 1.5Gig
The Syslog DB 216M
And the test part:
Cacti DB 9Gig
Flowview DB 182Gig (but not active at this time)
Syslog 2.5Gig (not active at this time)

But I had some issue with Maxscale, a test query was taking 12sec throurgh Maxscale and 0.3s directly on the server!
With some change on the config, I was able to reduce to 0.4s through Maxscale.
So I will look how it's gona do the update tonight.

[TheWitness]

You can run the IRR collection manually too. It's doing some real big UPSERTS. Checking that they are UPSERTS.

[TheWitness]

I see what the issue is. It's doing REPLACE INTO. That's a pretty expensive operation.

[TheWitness]

I just switched IRR database updates to UPSERTS. That should help with performance a little.