'update' in the title suggested to me that it is (new) SW functionality or regression is what causing the issue. Now with 'use the disk a lot' (I know nothing about boinc) it because like something we know for a decade already. Simple SD-card with their unknown and/or simple and not perfectly hard real-time under all circumstances maybe, will fail sooner than later if used 'a lot'.In the absence of a significant number other other reports I will be doing nothing other than replace the SD card with an SSD which I intended to do from day one anyway, some boinc tasks use the disk a lot.
What 'a lot' is, is also unknown as no-one developed a common intermediate blocklayer that counts I/O actions Maybe LVM, I am not aware. So the SD-card vendors, both the good and the grey-zone/fakers/criminals are very hard to compare. There are no numbers like from smartctl telling what has happened (the last 10 years or so).
If SSD, I would make sure you don't end-up in this situation: viewtopic.php?p=2353564#p2353186 Not able to let the storage device tell you what the level of wear-out is. Although that also might be due to old kernel, I don't know. Also look where the storage comes from, vendors have a simple way to hide info behind some random USB3 implementation.
Statistics: Posted by redvli — Mon Dec 15, 2025 11:31 am