[PATCH] nvme: reset retires after path failover

[Hannes Reinecke]

On 10/02/2017 04:43 PM, Keith Busch wrote:
> On Mon, Oct 02, 2017 at 04:35:03PM +0200, Johannes Thumshirn wrote:
>> When we do a path failover we should also reset the retry counter to 0
>> so we're not accidently failing I/O after a path failover.
>>
>> Signed-off-by: Johannes Thumshirn <jthumshirn at suse.de>
>> ---
>>  drivers/nvme/host/core.c | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
>> index 1ef11ca0fc5e..f1d8b239b7b6 100644
>> --- a/drivers/nvme/host/core.c
>> +++ b/drivers/nvme/host/core.c
>> @@ -118,6 +118,7 @@ static void nvme_failover_req(struct request *req)
>>  	blk_mq_end_request(req, 0);
>>  
>>  	nvme_reset_ctrl(ns->ctrl);
>> +	nvme_req(req)->retries = 0;
>>  	kblockd_schedule_work(&ns->head->requeue_work);
>>  }
> 
> Won't this make it possible to retry indefinitely? We've removed the retry
> time limit, so the only thing preventing inifinite retries on retryable
> errors is the max retry limit.
> 
Please do check my earlier patch series.
The idea is to first retry paths, and then failover to other paths.
Once all paths are exhausted we'll requeue the I/O in nvme_make_request().
So from that we shouldn't have any infinite retries.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		   Teamlead Storage & Networking
hare at suse.de			               +49 911 74053 688
SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton
HRB 21284 (AG Nürnberg)