Alot of gap in graphs

[saatchi]

My Technical support information of my cacti and server is as below. Cacti was working fine but suddenly the gaps and white lines in graphs started. It seems that spine couldn't check all 43866 datasources less than 1 minute. I tried to disable some devices and also I increased the number of threads and php servers and timed out values and max connection in mysql but it couldn't help me.
These lines are in cacti log file as errors :

MDPHP: Poller[0] ERROR: A DB Exec Failed!, Error:'1062', SQL:"insert into poller_output_rt (local_data_id, rrd_name, time, poller_id, output) values (15923, 'traffic_out', '2018-01-19 10:44:05', '19655', '1383806512434855')'

PHPSVR: Poller[0] ERROR: Input Expected, Script Server Terminating

SPINE: Poller[0] ERROR: Spine Timed Out While Processing Hosts Internal


I appreciate if anyone help me to solve this issue

Thank you


Cacti Version 0.8.8b
Cacti OS unix
SNMP Version NET-SNMP version: 5.7.2
RRDTool Version RRDTool 1.4.x
Hosts 2409
Graphs 24462
Data Sources Script/Command: 204
SNMP: 9401
SNMP Query: 17529
Script Query: 5
Script - Script Server (PHP): 101
Total: 27240
Poller Information
Interval 60
Type SPINE 0.8.8h Copyright 2002-2016 by The Cacti Group
Items Action[0]: 43569
Action[1]: 199
Action[2]: 98
Total: 43866
Concurrent Processes 24
Max Threads 70
PHP Servers 5
Script Timeout 100
Max OID 60
Last Run Statistics Time:58.6645 Method:spine Processes:24 Threads:70 Hosts:2360 HostsPerProcess:99 DataSources:43788 RRDsProcessed:0
PHP Information
PHP Version 5.6.31
PHP OS Linux
PHP uname Linux cacti 3.10.0-693.5.2.el7.x86_64 #1 SMP Fri Oct 20 20:32:50 UTC 2017 x86_64
PHP SNMP Installed
max_execution_time 500
memory_limit 1024M

[cigamit]

Cacti Version 0.8.8b
With your version of Cacti being seriously old (nearly 5 years) you probably won't have many people stepping up to troubleshoot.

I don't know your server specs, but this looks extremely high.
Concurrent Processes 24
Max Threads 70
You have to remember that increasing processes and threads will sometimes decrease performance, if your server can't handle the load. Its a bit of a balancing act. Generally I do 1 process per CPU core, and 12-15 threads per process (sometimes more depending latency to the devices). Your biggest issue is probably disk IO (just guessing, since you haven't posted any performance info), which means you should be using the boost plugin.