On Thu, Aug 07, 2014 at 02:43:45PM +0900, Hitoshi Mitake wrote: > At Wed, 6 Aug 2014 11:47:39 +0800, > Bingpeng Zhu wrote: > > > > zk_timeout is measured in milliseconds, so > > it should be divided by 1000 to seconds. > > > > Signed-off-by: Bingpeng Zhu <[email protected]> > > --- > > sheep/cluster/zookeeper.c | 2 +- > > 1 files changed, 1 insertions(+), 1 deletions(-) > > > > diff --git a/sheep/cluster/zookeeper.c b/sheep/cluster/zookeeper.c > > index 5c181e1..fea8258 100644 > > --- a/sheep/cluster/zookeeper.c > > +++ b/sheep/cluster/zookeeper.c > > @@ -971,7 +971,7 @@ static int zk_join(const struct sd_node *myself, > > rc = zk_node_exists(path); > > if (rc == ZOK) { > > sd_err("Previous zookeeper session exist, shoot myself. Please " > > - "wait for %d seconds to join me again.", zk_timeout); > > + "wait for %d seconds to join me again.", zk_timeout / > > 1000); > > # Ccing Ruoyu, because he is a maintainer of zookeeper components. > > If zk_timeout is less than 1000, the log will looks like "wait for 0 > seconds ..." and it would be confusing for users. How about changing > the message like this: "wait for %d milli seconds"?
DIV_ROUND_UP() for seconds looks better to me. Thanks Yuan -- sheepdog mailing list [email protected] http://lists.wpkg.org/mailman/listinfo/sheepdog
