If a process takes too long to respond to an IPC request, the master process may drop the connection, causing a broken pipe for the worker.
| Symptom | Likely fix | |---------|-------------| | Error after Zabbix restart | Remove stale IPC sockets | | Error on high load | Increase StartTrappers , StartPollers , check LimitNOFILE | | Error on specific host | Increase Timeout on that agent | | Error only with active checks | Match server/agent Timeout , check network stability | | Error + disk full | Clean /tmp or /var/run |
Too many concurrent UserParameter calls can exhaust file descriptors or process slots, causing pipe creation to fail. zabbix cannot write to ipc socket broken pipe upd
Add or modify the [Service] section:
| Primary Symptom | Most Likely Cause | Immediate Action | Long-term Solution | |----------------|-------------------|------------------|---------------------| | cannot write to IPC socket: Broken pipe + Too many open files | File descriptor exhaustion | Raise LimitNOFILE in systemd service | Increase OS limits; reduce concurrent items | | Error during high-discovery workloads | Discovery overload | Reduce StartDiscoverers temporarily | Spread discovery rules over longer intervals | | Error after upgrade to 7.4.x | Possible Zabbix bug or configuration mismatch | Roll back or check for patches | Review changelog; open Zabbix ticket | | Proxy database is down + broken pipe | Backend database connectivity | Restart database service | Configure proper DB persistence and reconnect | | Remote command execution failing | Agent disconnect during long-running commands | Adjust command timeout settings | Use asynchronous scripts or reduce command length | If a process takes too long to respond
If the Zabbix database is locked or slow, the server process might hang. When the frontend waits too long for the server to process a request, the socket connection times out and "breaks." Check for long-running SQL queries. Ensure the database has enough connections available. Common Fixes Increase Timeout Settings
A: Check the Zabbix JIRA issue tracker at support.zabbix.com. Search for "IPC socket broken pipe" filtered by your Zabbix version. Several tickets have been closed as "Fixed" or "Commercial support required," which may indicate a deeper underlying issue. When the frontend waits too long for the
By systematically increasing cache sizes, tuning process counts, fixing database performance, and ensuring proper OS limits, you can eliminate this error entirely. Remember: Zabbix is a distributed system even on a single server—treat its internal processes with the same respect you give external servers.
This is a Zabbix server-to-agent network issue. It’s a local OS‑level IPC (Inter‑Process Communication) failure on the agent host .