Zabbix version
- Zabbix Server: 6.4.21
- Zabbix Proxy: 6.4.21
- Database: SQLite
Problem description
We are facing an intermittent alert related to the Zabbix Proxy availability:
“Zabbix proxy last seen more than 600 seconds ago”
The alert occurs recurrently at specific time windows, affecting both development and homologation environments, and later resolves automatically without manual intervention.
The proxy eventually reconnects to the Zabbix Server and resumes normal operation.
Observed behavior
- The alert triggers when the proxy stops reporting for more than 600 seconds.
- After a period (ranging from a few minutes up to ~2 hours), the proxy reconnects and the alert is marked as RESOLVED.
- No manual restart is required.
- The issue is intermittent and time-based, not constant.
Proxy database details
The Zabbix Proxy is configured to use SQLite as its database:
DBName=/sqlite3/zabbix.db
Current database characteristics:
- SQLite database file size: ~500 MB
- Database stored on the root filesystem
- Disk space and inode usage are within safe limits
- No filesystem saturation or inode exhaustion detected
Relevant proxy configuration (zabbix_proxy.conf)
CacheSize=512M
StartDBSyncers=24
HistoryCacheSize=384M
HistoryIndexCacheSize=128M
Timeout=30
Configuration notes:
- Cache sizes were increased from defaults to reduce cache pressure.
- StartDBSyncers was increased to improve data synchronization throughput.
- The proxy host has ~8 GB of RAM available.
- No OOM events observed at OS level.
- CPU usage remains stable during the issue.
Hypothesis / suspicion
Given the intermittent nature and the fact that the proxy uses SQLite, we suspect one or more of the following:
- SQLite write lock/contention during peak history flush periods
- Proxy temporarily blocked while syncing history to SQLite
- Internal queue buildup causing delayed heartbeat to the server
- Limitations of SQLite under higher write concurrency
This may cause the proxy to miss heartbeats and trigger the “last seen more than 600 seconds ago” alert.
Questions for the community
1. Is this behavior expected when using SQLite with higher data volume and concurrency?
2. Are there known best practices or tuning recommendations for SQLite-based proxies in Zabbix 6.4?
3. Are there additional proxy parameters that could mitigate this intermittent disconnection?
Additional notes
- Network connectivity between proxy and server is stable.
- No packet loss or firewall drops observed.
- The issue affects multiple proxies with similar configuration.
- All proxies eventually recover without manual action.
- Zabbix Server: 6.4.21
- Zabbix Proxy: 6.4.21
- Database: SQLite
Problem description
We are facing an intermittent alert related to the Zabbix Proxy availability:
“Zabbix proxy last seen more than 600 seconds ago”
The alert occurs recurrently at specific time windows, affecting both development and homologation environments, and later resolves automatically without manual intervention.
The proxy eventually reconnects to the Zabbix Server and resumes normal operation.
Observed behavior
- The alert triggers when the proxy stops reporting for more than 600 seconds.
- After a period (ranging from a few minutes up to ~2 hours), the proxy reconnects and the alert is marked as RESOLVED.
- No manual restart is required.
- The issue is intermittent and time-based, not constant.
Proxy database details
The Zabbix Proxy is configured to use SQLite as its database:
DBName=/sqlite3/zabbix.db
Current database characteristics:
- SQLite database file size: ~500 MB
- Database stored on the root filesystem
- Disk space and inode usage are within safe limits
- No filesystem saturation or inode exhaustion detected
Relevant proxy configuration (zabbix_proxy.conf)
CacheSize=512M
StartDBSyncers=24
HistoryCacheSize=384M
HistoryIndexCacheSize=128M
Timeout=30
Configuration notes:
- Cache sizes were increased from defaults to reduce cache pressure.
- StartDBSyncers was increased to improve data synchronization throughput.
- The proxy host has ~8 GB of RAM available.
- No OOM events observed at OS level.
- CPU usage remains stable during the issue.
Hypothesis / suspicion
Given the intermittent nature and the fact that the proxy uses SQLite, we suspect one or more of the following:
- SQLite write lock/contention during peak history flush periods
- Proxy temporarily blocked while syncing history to SQLite
- Internal queue buildup causing delayed heartbeat to the server
- Limitations of SQLite under higher write concurrency
This may cause the proxy to miss heartbeats and trigger the “last seen more than 600 seconds ago” alert.
Questions for the community
1. Is this behavior expected when using SQLite with higher data volume and concurrency?
2. Are there known best practices or tuning recommendations for SQLite-based proxies in Zabbix 6.4?
3. Are there additional proxy parameters that could mitigate this intermittent disconnection?
Additional notes
- Network connectivity between proxy and server is stable.
- No packet loss or firewall drops observed.
- The issue affects multiple proxies with similar configuration.
- All proxies eventually recover without manual action.