[pve-devel] [PATCH qemu-server] Fix #1650: add content type filter to rescan

[Thomas Lamprecht]

On 7/3/18 8:28 PM, Dietmar Maurer wrote:
> I always tell users that is is dangerous to add the same storage 
> multiple times (using different names and options). This breaks very
> basic assumptions and locking will not work as expected.
> 

Hmm, shouldn't it be totally valid to have two storage entries from
the same backing storage with two complete different content types??
E.g., storeA with content images and storeB content backup.

There is no interference because they never want to do both the same
backup, or create a new disk at the same time.

> Also, the purpose of 'rescan' is to detect all stale volumes. So your 
> patch makes the whole thing pointless.
> 

No he doesn't makes it pointless. AFAIS, he only omits storages which
do not allow to store disk volumes, which is even a fix? No point in
adding a disk from a storage which the cluster should not but images
(i.e., volumes) on.

Seems not really wrong, IMO, and we use this for pve ceph, where the
same backing storage can get added twice, with and without KRBD to our
storage.cfg by our pool creator UI - one for CT which need KRBD and one
for VMs which normally do not want it.


> -1
> 
>> On July 3, 2018 at 5:41 PM Alwin Antreich <a.antreich at proxmox.com> wrote:
>>
>>
>> Unused disk(s) appeared after a rescan of storages. Especially shown
>> with ceph pools, where two storage entries are made, <storage>_ct and
>> <storage>_vm. The rescan method did include images from both storages.
>>
>> This patch filters any storage not containing the content type 'images'.
> 
> _______________________________________________
> pve-devel mailing list
> pve-devel at pve.proxmox.com
> https://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
>