On Sat, Feb 02, 2008 at 03:39:09PM +0800, Liang Zhen wrote: > Hi Kilian, > I think it's because o2iblnd uses fragmented RDMA by default(Max to > 256), so we have to set max_send_wr as (concurrent_send * (256 + 1)) > while creating QP by rdma_create_qp(), it takes a lot of resource and > can make a busy server out of memory sometime. > To resolve this problem, we have to use FMR to map fragmented buffer to > virtual contiguous I/O address, there will always be one fragment for > RDMA by this way. > Here is patch for this problem (using FMR in o2iblnd) > https://bugzilla.lustre.org/attachment.cgi?id=15144
This is an experimental patch - nodes with the patch applied are not interoperable with those without it. Please don't propagate the patch to production systems. Isaac _______________________________________________ Lustre-discuss mailing list [email protected] http://lists.lustre.org/mailman/listinfo/lustre-discuss
