On 05/14/2010 01:05 PM, Ian MacDonald wrote:
On Fri, 2010-05-14 at 11:46 -0500, Mike Christie wrote:
We had some issues with the initiator loosing connections with the
target in this new Karmic rootfs on iSCSI setup. The problem is
that
after some time the filesystem switches to a read-only mount
following
I/O errors after some activity on the initiator.
Do you have the log messages for that? Did you see any ping/nop
timeout
messages (for root on iscsi set node.conn[0].timeo.noop_out_interval
and
node.conn[0].timeo.noop_out_timeout to 0).
The logs have rolled over since the event unfortunately (just checked).
The output was minimalistic; two lines a few times.. and I recall a
generic timeout message. I think this is exactly what it was, but no
way to verify.
Unfortunately I don't want to change DNS again simply to cause the
problem.
That is fine, but I just want to make sure you are using the right iscsi
settings at least.
> multipath at the initiator (I also fear messing up my md devices). I
> assume multipath is enabled by default even with a single NIC from what
> I read here:
>
http://groups.google.com/group/open-iscsi/browse_thread/thread/5da7c08dd95211e6#
dm-multipath is not always enabled by default. It might be for some
distros, but I think normally it is not.
If you you have /sbin/multipath then run that to see what is setup. If
the scsi/iscsi devices are in a dm-multipath device that you are using
for root then you should be ok.
If you do not even have /sbin/multipath installed then you are not using
block layer multipath. In that case you would want the noop settings I
mention set to 0 (if you are using multipath set them lower but set
queue_if_no_path or no_path_retry to a high value).
>
> I also read this in the thread above, but do not know where to enable it
> since I have not /etc/multipath..
If you do not have the multipath.conf file then you probably do not have
multipath tools installed. I am not sure what the package is named in
ubuntu. In upstream it is called multipath-tools. In Red Hat based
distors it is named device-mapper-multipath-tools.
>
> "queue_if_no_path will internally queue IO until the path
> comes back or until the system runs out of memory or dies in some other
> way"
>
> but I know its there:
> [ 0.693148] device-mapper: multipath: version 1.1.0 loaded
> [ 0.693150] device-mapper: multipath round-robin: version 1.0.0
> loaded
If this is all you see, then you may not be using multipath. This just
indicates that the kernel modules needed for block layer multipath using
dm-multipath are loaded. It does not indicate that a dm-multipath device
has been created or is even being used. There is that separate
/sbin/multipath and and /sbin/multipathd programs that assemble
multipath devices by talking to those kernel modules.
--
You received this message because you are subscribed to the Google Groups
"open-iscsi" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/open-iscsi?hl=en.
