The automatic database export introduced in 1.35.1 leaked a database
connection on every run, so after a number of exports the controller's
connection pool was exhausted. If you are running 1.35.1 you should
upgrade or set AutoDbExport/Cron to "off" until you can.
The other fix is on the satellite side: if an integration outside of
LINSTOR left a volume's block device flagged read-only, deleting the
volume failed with "Failed to wipeFs" and the volume was never removed.
The satellite now clears the read-only flag before wiping the device.
[1.35.2] - 2026-09-14
Fixed
- Satellite: Clear the read-only flag of a block device before wiping
it, so that deleting a volume no longer fails with "Failed to wipeFs"
if an external user outside of LINSTOR left the device flagged
read-only - Controller: Fixed a database connection leak in the automatic
database export, which exhausted the connection pool after a number of
exports