#4 gives the server fixed thresholds for disk headroom: it warns when the data volume has under 100 MB free and calls it critical under 50 MB, and a write that would leave less than the reserve (the larger of 50 MB and the index database's own size) is refused. Those numbers suit a markdown vault on a Railway volume, which is what this server is for.
If anyone runs it somewhere the numbers are wrong — a very large attachment-heavy vault, or a host where 100 MB is nothing — the fix is two environment variables, say DISK_WARN_BYTES and DISK_CRITICAL_BYTES, read at boot with the defaults above. Not doing it until someone asks: an override nobody uses is a setting to document and test forever.
Refs #4.
#4 gives the server fixed thresholds for disk headroom: it warns when the data volume has under 100 MB free and calls it critical under 50 MB, and a write that would leave less than the reserve (the larger of 50 MB and the index database's own size) is refused. Those numbers suit a markdown vault on a Railway volume, which is what this server is for.
If anyone runs it somewhere the numbers are wrong — a very large attachment-heavy vault, or a host where 100 MB is nothing — the fix is two environment variables, say
DISK_WARN_BYTESandDISK_CRITICAL_BYTES, read at boot with the defaults above. Not doing it until someone asks: an override nobody uses is a setting to document and test forever.Refs #4.