If your poller processes are constantly maxed out, they will queue up data faster than the syncers can process it, leading to socket timeouts. Open your zabbix_server.conf or zabbix_proxy.conf . Increase the number of pollers to distribute the load:

PID in your logs. If you see it restarting frequently, it may be crashing due to a specific malformed item or excessive load Increase the for the specific service to get more detail: zabbix_server -R log_level_increase="preprocessing manager" 4. Verification & Clean Restart

If the primary database (PostgreSQL or MariaDB/MySQL) slows down or hits maximum connections, Zabbix history syncers hang. The preprocessing queue backs up until it times out and forcefully kills internal connections. Step-by-Step Troubleshooting and Resolutions 1. Increase Linux File Descriptor Limits ( ulimit )

The "cannot write to IPC socket: broken pipe" error in Zabbix is essentially a symptom of a process unable to hand off its workload to the next step in the monitoring pipeline. It often points to a chain reaction: To effectively resolve this:

The StartPreprocessors parameter controls how many concurrent preprocessing worker processes Zabbix runs. When set too high, these workers can overwhelm the internal queue buffers and cause IPC pipes to break. As a documented workaround, setting StartPreprocessors=1 can temporarily stabilize the system until the root cause is addressed.

Zabbix relies heavily on sockets. These are internal pathways that allow the main Zabbix server process to talk to pollers, trappers, and the database syncers.

sudo nano /etc/security/limits.conf

The Zabbix error "cannot write to IPC socket: Broken pipe" typically indicates that a communication channel between internal Zabbix processes (like the server and its preprocessing manager) has been severed. This often occurs when the receiving end of a socket closes unexpectedly while another process is still trying to send data, frequently due to high system load or configuration limits. Common Causes of Broken Pipe Errors

Look for lines indicating zabbix_server was terminated by the OOM-killer. Step 2: Analyze Zabbix Log Timestamps

StartDiscoverers=10

This is the most common and often the most effective fix.