Force tcp instead of udp

[Lejooohn]

Hi there !

a long time ago we could use this trick : tcp:hostname:port to force using tcp instead of udp.

We using now the latest version of cacti (and spine) and this trick doesn't work anymore.

How can we do to force tcp connexion now? (we dont want use tcp connexion on all of ours devices, just few of them).

Regards

[bmfmancini]

you can set the availability to TCP ping and set the port you want to check against

[TheWitness]

That's not it @bmfmancini. Looks like a spine issue.

[Lejooohn]

you can set the availability to TCP ping and set the port you want to check against

TCP ping works yes but i need to keep the device detection using "SNMP uptime" that's why i asking my question here :)

That's not it @bmfmancini. Looks like a spine issue.

Can i do something or it's an issue need to be resolve by the maintainers?

best regards,

[TheWitness]

Run

./spine -V 5 -R -S -f id -l id -mibs

Replace "id" with the device id in question. Send output.

[Lejooohn]

-V 5 -R -S -f id -l id -mibs

Hi @TheWitness

When i use the -mibs parameters :

/usr/local/spine/bin/spine -V 5 -R -S -f 320 -l 320 -mibs

2021/01/08 12:47:56 - SPINE: Poller[1] PID[8265] ERROR: -mibs is an unknown command-line parameter (Spine init)

without and when the device is configured to use the hostname like we did before :

/usr/local/spine/bin/spine -V 5 -R -S -f 320 -l 320
SPINE: Using spine config file [/etc/spine.conf]
2021/01/08 12:46:39 - SPINE: Poller[1] PID[7923] DEBUG: The path_php_server variable is /opt/cacti-1.2.16/script_server.php
2021/01/08 12:46:39 - SPINE: Poller[1] PID[7923] DEBUG: The path_cactilog variable is /var/www/html/cacti/log/cacti.log
DEBUG: The log_destination variable is 4 (STDOUT)
DEBUG: The path_php variable is /usr/bin/php
DEBUG: The availability_method variable is 4
DEBUG: The ping_recovery_count variable is 2
DEBUG: The ping_failure_count variable is 2
DEBUG: The ping_method variable is 1
DEBUG: The ping_retries variable is 3
DEBUG: The ping_timeout variable is 4000
DEBUG: The snmp_retries variable is 10
DEBUG: The log_perror variable is 1
DEBUG: The log_pwarn variable is 1
DEBUG: The boost_redirect variable is 0
DEBUG: The boost_rrd_update_enable variable is 0
DEBUG: The log_pstats variable is 1
DEBUG: The threads variable is 40
DEBUG: The polling interval is 300 seconds
DEBUG: The number of concurrent processes is 2
DEBUG: The script timeout is 10
DEBUG: The selective_device_debug variable is 1009,1033
DEBUG: The spine_log_level variable is 0
DEBUG: The number of php script servers to run is 10
DEBUG: StartDevice='320', EndDevice='320', TotalPHPScripts='0
DEBUG: The PHP Script Server is Not Required
DEBUG: The Maximum SNMP OID Get Size is 1
Selective Debug Devices 1009,1033
Version 1.2.16 starting
DEBUG: MySQL is Thread Safe!
DEBUG: Spine running as 0 UID, 0 EUID
DEBUG: Spine is running as root.
DEBUG: Spine has got ICMP
SPINE: Initializing Net-SNMP API
DEBUG: Issues with SNMP Header Version information, assuming old version of Net-SNMP.
SPINE: Initializing PHP Script Server(s)
NOTE: Spine will support multithread device polling.
DEBUG: Initial Value of Active Threads is 0
DEBUG: Valid Thread to be Created
DEBUG: In Poller, About to Start Polling of Device for Device ID 0
SPINE: Active Threads is 1, Pending is 1
DEBUG: Valid Thread to be Created
DEBUG: In Poller, About to Start Polling of Device for Device ID 320
SPINE: Active Threads is 2, Pending is 2
Device[320] INFO: SNMP Device 'tcp:hostname:1161' has timeout 3000000 (3000), retries 10
ERROR: Problem initializing SNMP session 'tcp:hostname:1161
Device[0] HT[1] Total Time: 0.0088 Seconds
Device[0] HT[1] DEBUG: HOST COMPLETE: About to Exit Device Polling Thread Function
DEBUG: The Value of Active Threads is 1 for Device ID 0
POLLER: Active Threads is 1, Pending is 1
WARNING: Unable to determine address info for tcp:hostname.com:1161 (Name or service not known)
Device[320] DEBUG: Entering SNMP Ping
Device[320] PING Result: Ping not performed due to setting.
Device[320] HT[1] NOTE: There are '18' Polling Items for this Device
Device[320] HT[1] Total Time: 0.018 Seconds
Device[320] HT[1] DEBUG: HOST COMPLETE: About to Exit Device Polling Thread Function
DEBUG: The Value of Active Threads is 0 for Device ID 320
POLLER: Active Threads is 0, Pending is 0
SPINE: The Final Value of Threads is 0
DEBUG: Thread Cleanup Complete
DEBUG: PHP Script Server Pipes Closed
DEBUG: Allocated Variable Memory Freed
DEBUG: MYSQL Free & Close Completed
DEBUG: Net-SNMP Close Completed
Time: 0.1406 s, Threads: 40, Devices: 2

however the fqdn "hostname" exist but perhaps because i set "tcp:" and the port in the hostname the spine think it's a part of the hostname and not a simply parameters.

[TheWitness]

Okay, the issue is clear now. There was a change to move away from gethostbyname() to getaddrinfo(), and the author was unaware of this config option.

[Lejooohn]

Okay, the issue is clear now. There was a change to move away from gethostbyname() to getaddrinfo(), and the author was unaware of this config option.

Arf thks for the explainations @TheWitness . So there is no way to recover this trick or an other way to configure a SNMP for a device with TCP check?

Regards,

[TheWitness]

It's a spine bug. See about getting it fixed. Moving this issue to Spine bug tracker though.

[netniV]

I have made a change for this, but I think it may need verifying and extending. You should review it @TheWitness as it may change/break some functionality.