19.03.2019 14:01, Kevin Wolf wrote: > Am 19.03.2019 um 10:24 hat Vladimir Sementsov-Ogievskiy geschrieben: >> Job (especially mirror) may call block_job_error_action several >> times before actual pause if it has several in-flight requests. >> >> block_job_error_action will call job_pause more than once in this case, >> which lead to following block-job-resume qmp command can't actually >> resume the job. >> >> Fix it by do not increase pause level in block_job_error_action if >> user_paused already set. >> >> Signed-off-by: Vladimir Sementsov-Ogievskiy <vsement...@virtuozzo.com> > > Thanks, applied to the block branch. (But please follow up with a test > case anyway.) > >> Hi all! >> >> I have a reproducer, however, I don't know how to make a test from it, >> as I don't know a way to create a size-limited directory without root >> access. > > Maybe we can use the 'size' option of the raw format driver below a > qcow2 layer to achieve something like this? You can update the size at > runtime using 'reopen -o size=...' in HMP qemu-io.
Cool that works, thanks! > > file <- raw [size=11M] <- qcow2 [size=1G] > > This should return an error after writing the first 11M (including > metadata), and once you resize the raw layer, you should be able to > resume the job. > > Kevin > -- Best regards, Vladimir
