25.3.5
Published: 9 July 2026
Bug fixes
Fixed an issue where Oracle database checks intermittently failed with "Closed Connection" errors after the database server closed an idle monitoring connection, for example, due to Oracle profile idle time limits or Resource Manager policies. Avantra now detects the stale connection and reconnects automatically.
Fixed an issue where the JavaSSLCertificatesValidity check failed with a ClassNotFoundException after upgrading SAP NetWeaver AS Java to 7.50 SPS 34 or later, preventing the check from reporting certificate validity.
Fixed an issue where an agent downloading an update package through a gateway threw a NullPointerException while decoding the response headers, preventing the agent from updating.
Fixed an issue in clustered (VCS) environments where an agent on a cluster node that did not currently own a virtual IP still monitored the SAP instances hosted on that IP. Because the virtual IP was not bound locally, the passive node could not reach the instances and reported the SAP system as DOWN, overwriting the correct status from the active node. Avantra no longer runs detection or checks for instances whose virtual server is not active on the local node.
Fixed an issue where SAP Cloud ALM checks could use the wrong metric ID because the SAP system type was not taken into account, causing checks to return no data for systems such as SAP S/4HANA. Metric IDs are now resolved per system type. To pick up the system type for an existing system, remap the system.
Fixed an issue on large agent fleets where the agent did not persist its change-detection state, so every agent restart caused all configuration data (such as profiles, SAP clients, and database parameters) to be re-sent to the master as changes. At scale, these bursts overloaded the master, producing large numbers of AgentAlive CRITICAL alerts and causing monitored systems to intermittently show as UNKNOWN.