Hello again,
On 06/24/2014 07:24 AM, Martin George wrote:
On 6/24/2014 4:11 AM, Hans van Kranenburg wrote:
On 06/23/2014 06:31 PM, Hans van Kranenburg wrote:
Questions left:
- Is it wanted to have the linux kernel multipathing fail an iop
instead of retry on receiving the combination of a medium error and
additional code incompatible medium installed?
Well, it would have been ideal if the Linux kernel had given up for this
MEDIUM ERROR - INCOMPATIBLE MEDIUM INSTALLED ASC as well, instead of
retrying on all available paths (since retrying would only end up
hitting the same error again, given that this is a device error).
Yeah, that's true. In that case, even if the actual medium error would
be a bug, which is the case here, it would fail a single path, then fail
the iop and then after the direct-io checker would come along, the path
would be enabled again.
I'll leave this as an exercise for myself to create a kernel patch that
would do this. Seems doable, the only problem is that I need a
development environment that simulates exactly this behaviour. I can
obviously not use our production netapp system for this. :)
- Now I'm left with my broken NetApp, and I'd like to start using
UNMAP on it... Any comments from netapp people reading this? There must
be some reason why this is happening, and only on this specific lun, and
not on the test lun, or on several of the other NetApp filer we use.
Yes, the NetApp controllers are returning this MEDIUM ERROR check
condition for some reason. I'd suggest you open a NetApp support ticket
for tracking this.
And so I did. The answer from NetApp support is:
"Since the UNMAP command is an host OS related command, it is not
something that we implement in explicitly in Data ONTAP. The
functionality to support the command is something we do have to test
however, and any changes or additions to support it are done in new
releases. This is why you are able to get some functionality out of the
command in 8.1.2. However, it's not fully certified in Data ONTAP until
8.1.3."
So, using UNMAP might work in ONTAP versions before 8.1.3, but it does
not have an official approval stamp by NetApp.
Also see:
- https://kb.netapp.com/support/index?page=content&id=3013806
- https://kb.netapp.com/support/index?page=content&id=3013991
So, although this functionality is reported as 'working' in many cases,
it seems that NetApp itself has fixed some bugs in 8.1.3 and only then
officially started supporting using it.
Right now, our idea is to do some upgrades from random 8.1 versions we
run to the latest maintenance version of ONTAP 8.1 (8.1.4P1), which is
not a bad idea in any case.
Anyway, I'm still typing this in a debian bug report. I know this is
actually not a debian bug at all anymore, but since I guess that whoever
encounters the same issue and starts to look for a solution on the
internet ends up reading this page, I'd like to use this bug report to
document my experiences.
If Bill (OP) agrees, I'll close it as soon as I found a solution for the
problem, documenting it, whatever that is.
--
Hans van Kranenburg - System / Network Engineer
T +31 (0)10 2760434 | hans.van.kranenb...@mendix.com | www.mendix.com
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
