Disk buffer should be a ring-buffer

[megastef]

Logagent buffers logs (by default 10000 bulk requests with a max. size of 1000 logs) on disk when shipping via HTTPS/HTTP fails. If such a problem remains longer, logagent would stop to buffer logs on disk. I think most users would prefer to drop the oldest file to keep most recent logs in the (always limited) disk buffer.

Current workarounds:

1. If disk space is not an issue, one could increase the limit to a higher value e.g. 50k bulk files.
   export LOGSENE_MAX_STORED_REQUESTS=50000
2. Run a chron job to delete oldest files form LOGSENE_TMP_DIR. Some potential solutions are here https://stackoverflow.com/questions/2960022/shell-script-to-count-files-then-remove-oldest-files

See https://github.com/sematext/logsene-js#environment-variables for further details of logsene-js settings.

[megastef]

Should be solved with the update to latest logsene-js e9ca4be - logsene-js deletes the oldest file, when the maximum number of stored files is reached - so the disk buffer behaves like a ring-buffer.