On 15:09 Tue 22 Oct, Shawn Chen wrote:
Hi,changcheng, I wanna know if this problem has been
resolved since I
met this problem too~
wish your reply, thanks.
I think there's one serious memory leak in
Ceph/RDMA/iWARP implementation: The posted allocated outstanding work-request were not
reclaimed.
The problem should be fixed in below PR:
https://github.com/ceph/ceph/pull/29947
You can use below master version to do verfication:
commit 9068510305
--Thanks
Changcheng
>
> ---------- Forwarded message ---------
> 发件人: Liu, Changcheng <changcheng.liu(a)intel.com>
> Date: 2019年5月23日周四 下午1:10
> Subject: Re: msg/async/rdma: out of buffer/memory
> To: Roman Penyaev <rpenyaev(a)suse.de>
> Cc: <ceph-devel(a)vger.kernel.org>rg>, <ceph-devel-owner(a)vger.kernel.org>
>
>
> Hi Roman,
> I'll check it and feedback to you later. (Something wrong with
> servers, I have to fix it first)
>
> B.R.
> Changcheng