On 02.12.20 01:21, Micka wrote: > I've a directory that is watched for the event write-close. > When I get this event I assume that the file is received totally. > Is it correct? No, an open FD *will* eventually be closed, by the OS when the process terminates if need be. Unless the machine *crashes* in mid-operation, of course ... We do a lot of such transfers, and the (our) gold standard, if possible, is to do an upload-then-rename under client control, with the further processing stages on the server requiring a "non-temp name" match. (Since you didn't specify how exactly you transfer the data: This can be done with sftp and ssh(*), but not scp.) (*) E.g., a la "uuencode tempfile < sourcefile | ssh user@host "uudecode && mv tempfile finalfile", not that I'd *recommend* such an approach ... Regards -- Jochen Bern Systemingenieur Binect GmbH
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ openssh-unix-dev mailing list openssh-unix-dev@xxxxxxxxxxx https://lists.mindrot.org/mailman/listinfo/openssh-unix-dev