[sheepdog-users] [sheepdog] filesystem errors when using sheepdog

[Valerio Pachera]

2015-07-09 14:54 GMT+02:00 Maxim Terleckiy <max.terleckiy at gmail.com>:

> As I remember discard is not supported by virtio bus, so you need to use
> virtio-scsi instead. In that case it will be necessary to add virtio-scsi
> controller to vm definition.
>
> I had similar problems with discard and missing blocks at 0.8.0, so I
> think it is some kind of long last bug. Additionally I had problem with
> virtio-scsi attached disks in case of reconnection to cluster after cluster
> unavalability. With simple virtio when io is stuck, vm is just waiting for
> cluster responding, trying to reconnect and so on. With virtio-scsi my vms
> were always diying with kernel panics even during short periods of cluster
> unavailibility.
>

Thank you for sharing that.
BTW, I have not clear if the issue is related to qemu block dirver, virtio
driver itself or sheep daemon.

PS:
FITRIM is a mounted filesystem feature to discard (or "trim") blocks which
are not in use by the filesystem. This is useful for solid-state drives
(SSDs) and thinly-provisioned storage.

There are many discussion about trim.
I alse heard recently that trim is managed directly by the disk controller.
So, I wonder if discard may be useful or not at all.
Thank you for sharing that.
BTW, I
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.wpkg.org/pipermail/sheepdog-users/attachments/20150709/4c6218f1/attachment.html>