Hi, On Tue, 21 Sept 2021 at 09:30, freoone <freoone@xxxxxxxxx> wrote: > > > I'm trying to run fio to do some random reads with 2 threads > (numjobs=2). I'd like to record iolog for both threads in separate > files. > The manual in the 'write_iolog' section says: > "Specify a separate file for each job, otherwise the iologs will be > interspersed and the file may be corrupt." > and later: > "You can specify a number of files by separating the names with a ‘:’ > character." The text "you can specify a number of files by separating the names with a ‘:’ colon" only applies to the explicit filename option (https://fio.readthedocs.io/en/latest/fio_man.html#cmdoption-arg-filename ) as opposed to other options (like write_iolog) that happen be filenames. > So I'm running the following command: > ./fio --name=test2jobs --rw=randread --filename=file.fio --size=10M > --numjobs=2 --write_iolog=log1:log2 > but instead of two files "log1" and "log2" I get a single file > "log1:log2" containing IOs from both threads. > Is this a bug in fio or is there some misconception on my part? To get what you describe you have to avoid usage of numjobs. For example if I leave make most of the options as global then this works: ./fio --rw=randread --filename=file.fio --size=10M --name=testjob1 --write_iolog=log1 --name=testjob2 --write_iolog=log2 -- Sitsofe
