[PATCH] nvme: rdma/tcp: call nvme_mpath_stop() from reconnect workqueue

Martin Wilck mwilck at suse.com
Fri Apr 23 10:33:24 BST 2021


On Fri, 2021-04-23 at 16:35 +0800, Chao Leng wrote:
> 
> If ana_work is running, this may cause wait for long time, because
> nvme_get_log may time out(default 60s). If work with multipathing,
> it will cause fail over delay, service will pause long time.
> It is not we expected.
> We just need to do this before reconnecting, so move it until
> calling nvme_rdma_reconnect_or_remove.
> Like this:
> ...

Thanks, that makes sense. I'll look into it.

Martin





More information about the Linux-nvme mailing list