Spine not updating data for a device after upgrade to 1.2.10

[camile]

Hello,

I'm upgrading my cacti server from 0.8.8b (with package installed from ubuntu 14.04 repositories) to 1.2.10 (with package provided with Ubuntu 20.04).

The process for the upgrade was:

• I copied the database from one server to the new one (mysql is in the same host)

• I installed and configured the new cacti, connected to the new database

• I ran the installation script which updated the database

The problem I'm having in the updated server is that I have a few windows servers that aren't monitored (almost all my devices are network equipment, not servers).

Trying to debug the problem, I have disabled all devices except one of the windows servers. I have also tried to create this device from the beginning. But the problem is still there, none of the are being monitored.

Then I have tried to change from spine to cmd.php poller. In this case, using cmd.php both devices (the imported and the new one) are correctly monitored, but if I change back to spine, then monitoring is stopped again.

I attach screenshots with device, datasource and graph config for the imported and the new device. In the graphs, the ranges with data are the ones I have tried using the cmd.php poller.

I have tried to debug the problem, but I couldn't find any relevant information. Can anybody help me? Thanks.

[camile]

I don't know if this could help...

At this moment, I have disabled all devices except the old and new I'm trying to debug. In each of these devices I have just one graph.

But cacti is loggin:

2021/02/25 12:27:02 - SYSTEM STATS: Time:1.2252 Method:spine Processes:1 Threads:15 Hosts:2 HostsPerProcess:2 DataSources:0 RRDsProcessed:0

I have checked that related devices and data sources are enabled.

[camile]

I have created a new device to continue debugging. Then I have manually run spine in debug mode (-V 10). I can see that it makes queries like:

DEVDBG: SQL:SELECT snmp_port, count(snmp_port) FROM poller_item WHERE host_id=673 AND rrd_next_step <=0 AND poller_id=1 GROUP BY snmp_port
DEVDBG: SQL:SELECT action, hostname, snmp_community, snmp_version, snmp_username, snmp_password, rrd_name, rrd_path, arg1, arg2, arg3, local_data_id, rrd_num, snmp_port, snmp_timeout, snmp_auth_protocol, snmp_priv_passphrase, snmp_priv_protocol, snmp_context, snmp_engine_id FROM poller_item WHERE host_id=673 AND rrd_next_step <=0 AND poller_id=1 ORDER by snmp_port

so I'm assuming that spine looks in poller_item table for element with rrd_next_step filed less or equal to 0 and then run the related queries.

Then, I have enabled other devices and I can see that every time spine is run (every minute), the rrd_next_step for the working device is decrease by 60 (1 minute) and when its value is <= then it is queried.

But for my problematic devices this field is never updated, so it never reaches a value <= 0.

I have also tried to manually updated the value to 0, then the query is made in the next spine run and the rrd_next_step is updated to 300, but it is not updated by later spine runs.

[TheWitness]

Try updating to 1.2.17. This is likely happening as the SysDecription has non-ascii characters so it output's the string as hex. Does that sound right? Does it have non-ascii characters, like characters from a foreign laguage?

[camile]

Yes, it has a SysDescription like that. I'm going to try 1.2.17.

[camile]

Sorry, but last version in the web is 1.2.16. Where can I download 1.2.17?

[Osiris]

git clone -b 1.2.x https://github.com/cacti/cacti.git

[camile]

I updated to 1.2.16 and it seems to fix the isuue. Should I update to 1.2.17?