Request timeout seen with NVMEoF TCP

Potnuri Bharat Teja bharat at chelsio.com
Mon Dec 21 02:50:53 EST 2020


On Friday, December 12/18/20, 2020 at 17:35:01 -0800, Sagi Grimberg wrote:
> 
> > > Potnuri, does this patch solves what you are seeing?
> > 
> > Hi Sagi,
> > Below patch works fine. I have it running all night with out any issues.
> > Thanks.
> 
> Thanks Potnuri, can you try the patch without the nvme_tcp_io_work
> portion? I think the issue can be resolved with just adding
> nvme_tcp_send_all..
> 
Yes. Issue is resolved with just nvme_tcp_send_all() change:
Test runs fine for me.
Thanks.

> Thinking more about what Mark mentioned and it is correct
> that io_work needs to reschedule itself if it wasn't able
> to acquire the send_mutex hence I am not sure it is needed.
> 



More information about the Linux-nvme mailing list