Rado S schreef:
as already pointed out by others, with LOGABSTRACT=all you get
multiple write-line operations per delivery logged, but concurrent
deliveries don't lock the logfile, so they can get mixed up.
On the other hand, if instead of using multi write-line operation
logging you use LOGABSTRACT=no single write-line operation per LOG
or TRAP (the written line containing \n: the output is the same in
both cases, just number of operations reduced to 1), then people
believe (or not, in my case ;) that this is corruption safe.
Even though this might be a single write-line operation, still
each char has to be written. Is there some locking active to avoid
char race condition between 2 write-line operations?
Call it a single-buffer write then. Writing is not done by character (by
any modern OS).
An OS-buffer can be for example 2kB (but also smaller, or (much) bigger)
but a single write can handle multiple OS-buffers.
--
Groet, Ruud
____________________________________________________________
procmail mailing list Procmail homepage: http://www.procmail.org/
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail