On 25/06/2024 20:04, J. Hannken-Illjes wrote:
Good to know that it won't actually mess the filesystem integrity up. I've seen it precisely once and I've done 100s of bulk pkgsrc builds on this system and even more OS builds.While flushing dirty blocks on a device node new dirty blocks appeared. In theory this should not happen -- but the vflush will retry anyway. If you see it one or two times there is no real problem.
I might need to investigate why dirty block flushing is taking so long as it could indicate an SSD that's on its way out (saw something similar on another system).
Could having the 'discard' option on impact things? I know TRIM operations can sometimes be slow.
Mike