On Tue, Jun 5, 2012 at 3:59 PM, Pekka Enberg <penberg@xxxxxxxxxx> wrote: > On Mon, 4 Jun 2012, Asias He wrote: > >> The queue size for virtio_blk is 256 and AIO_MAX is 32, we might be >> short of available aio events if guest issues > 32 requests >> simultaneously. Following error is observed when guest running stressed >> I/O workload. >> >> Info: disk_image__read error: total=-11 >> >> To fix this, let's increase the aio events limit. >> >> Signed-off-by: Asias He <asias.hejun@xxxxxxxxx> >> --- >> tools/kvm/disk/core.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/tools/kvm/disk/core.c b/tools/kvm/disk/core.c >> index ed338e7..d1d2d59 100644 >> --- a/tools/kvm/disk/core.c >> +++ b/tools/kvm/disk/core.c >> @@ -6,7 +6,7 @@ >> #include <sys/eventfd.h> >> #include <sys/poll.h> >> >> -#define AIO_MAX 32 >> +#define AIO_MAX 256 >> >> int debug_iodelay; > > Is there any downside to bumping up the limit? Ingo? We can have 0x10000 aio requests by default which is way much larger than 256. And we can also increase this limit through /proc/sys/fs/aio-max-nr. ------> fs/aio.c unsigned long aio_max_nr = 0x10000; /* system wide maximum number of aio requests */ <------ -- Asias He -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
