The agent manager crashes frequently.
When checking the settings, the cache folder under \state\default\cache contain files that are filling up the disk.
The upstream_data.cache file continues growing and it doesn't stop. After have cleaned the cache files and a Fglam restart, the issue still persist.
In the MongoDB, Postgresql, MySQL, and Cassandra database cartridges the following information can be seen in the agent.manifest files: max-queue-size and max-disk-space. Some of the cartridges has configured max-disk-queue = "-1". This configuration allows the Fglam to grow the cache files with no limits.
The rest of database cartridges (SQL Server, Oracle, DB2 and Sybase) do not have this configuration (max-queue-size and max-disk-space) in their agent.manifest file.
By default, the fglam.config.xml file is the one that contains the information for max-queue-size and max-disk-space.
The agent.manifest file for these cartridges override the settings configured in fglam.config.xml.
In consequence, the cache file will grow unlimited and the agent manager could crash.
WORKAROUND:
Please only apply this change if the MySQL, Cassandra, MongoDB, and PostgreSQL agent has the max-queue-size setting in the agent manifest file (versions below 5.9.3.10)
The agent.manifest files can be located on:
Note: this workaround needs to be applied in each agent manager affected.
STATUS:
Beginning with the 5.9.3.10 and higher cartridges the agent manifest in these cartridges no longer contains a max-queue-size value
