[sheepdog-users] Storage loss due to error

[Hitoshi Mitake]

Hi Vedvyas,

Thanks for reporting. I think the main source of the problem is fixed
timeout of sheepdog. e.g.
https://github.com/sheepdog/sheepdog/blob/master/include/net.h#L17

The fixed timeout value is unrealistic so we want to fix it (e.g. preparing
configuration mechanism). Could you try with an expanded timeout at first?
If the problem can be solved, we should work on the configuration mechanism.

Thanks,
Hitoshi


On Mon, May 9, 2016 at 1:50 AM, Vedvyas shanbhogue <vedvyas13686 at gmail.com>
wrote:

> Running a stress test where I have a VDI conneted over ISCSI. I have a
> script on the client that copies a 4G .iso file to a new folder in the VDI
> and then runs a SHA1 check on the copied .iso and then deletes the .iso.
> After a few iterations of this script I sometimes see the storage is lost -
> the client times out and aborts. In sheepdog logs I see following messages.
> Any suggestions/hints on debugging this failure?
>
> May 08 12:34:14   WARN [gway 5018] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:34:14   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:26   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:31   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:36   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:41   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:46   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:35:51   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
> May 08 12:36:02   WARN [gway 5019] wait_forward_request(390) poll timeout
> 1, disks of some nodes or network is busy. Going to poll-wait again
>
>
>
>
> --
> sheepdog-users mailing lists
> sheepdog-users at lists.wpkg.org
> https://lists.wpkg.org/mailman/listinfo/sheepdog-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.wpkg.org/pipermail/sheepdog-users/attachments/20160513/e8434f91/attachment.html>