Subject: Re: Question: various bugs in sync()?
To: None <firstname.lastname@example.org>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
Date: 01/15/1999 10:57:46
Some of us BOFs who used VAXes still type "sync" three times before
shutting a system down. It used to be that typing "sync" on a
hardcopy console was long enough for a massbus disk to complete the
writes scheduled (but not completed) by sync().
Intuitively, on the first sync, one could hear the disk activity as
cached data went to disk; on the second sync, any updated metadata
from the previous syncs got written; and on the last sync the disk
should be more-or-less silent.
Hm. i just tried a "sync" on a large Solaris 2.6 box, and it took
several seconds before the shell prompt came back....