On 12/02/2011 06:20 AM, Jeremy Sanders wrote:
> 
> Mike Christie wrote:
> 
>> Not exactly. iscsid.conf is used as the default settings for the
>> node/portal records. When you run the iscsiadm discovery/new command
>> then iscsiadm will create records in /etc/iscsi/nodes or
>> /var/lib/iscsi/nodes (depends on the version) for each portal
>> found/setup. When you run iscsiadm -m node -o show you are seeing those
>> values in the db. When the iscsi service starts (when we login to a
>> target), iscsiadm reads the db and uses those settings for each session.
>> So some of those settings might be what we ask the target to use but we
>> might end up negotiating some other value with the target.
> 
> Thanks for the helpful response. What about the iscsistart program which 
> dracut uses to get the initial connection? I assume this doesn't set these 
> settings, but does the iscsid which starts afterwards affect the connection 
> created by iscsistart?

Yes. iscsistart just uses some hard coded defaults. When iscsid starts
from the real root, it will read the db (/var/lib/iscsi/nodes in F16)
and force the current session to logout and then login with the new
settings.

> 
>>>> On the F16 system is iscsid running?
>>>
>>> Apparently (systemctl reports the service as running):
>>> root       795  0.0  0.0  12908   656 ?        Ss   Nov30   0:01 iscsid
>>> root       796  0.0  0.1  15796  5576 ?        S<Ls Nov30   0:04 iscsid
>>>
>>
>> So it is up. Could you send the /var/log/messages for when this happens?
> 
> It's broken again now The logs currently show:
> 
> Dec  2 09:58:56 xs2 kernel: [   14.836794] Loading iSCSI transport class 
> v2.0-870.
> Dec  2 09:58:56 xs2 kernel: [   14.838084] iscsi: registered transport (tcp)
> Dec  2 09:58:56 xs2 kernel: [   14.903442] iscsistart (446): 
> /proc/446/oom_adj is deprecated, please use /proc/446/oom_score_adj instead.
> Dec  2 09:58:56 xs2 kernel: [   16.154330] scsi4 : iSCSI Initiator over 
> TCP/IP
> Dec  2 09:58:56 xs2 kernel: [   16.420384] scsi 4:0:0:0: RAID              
> IET      Controller       0001 PQ: 0 ANSI: 5
> Dec  2 09:58:56 xs2 kernel: [   16.420685] scsi 4:0:0:0: Attached scsi 
> generic sg0 type 12
> Dec  2 09:58:56 xs2 kernel: [   16.421884] scsi 4:0:0:1: Direct-Access     
> IET      VIRTUAL-DISK     0001 PQ: 0 ANSI: 5
> Dec  2 09:58:56 xs2 kernel: [   16.422189] sd 4:0:0:1: Attached scsi generic 
> sg1 type 0
> Dec  2 09:58:56 xs2 kernel: [   16.423111] sd 4:0:0:1: [sda] 33204224 512-
> byte logical blocks: (17.0 GB/15.8 GiB)
> Dec  2 09:58:56 xs2 kernel: [   16.425964] sd 4:0:0:1: [sda] Write Protect 
> is off
> Dec  2 09:58:56 xs2 kernel: [   16.426701] sd 4:0:0:1: [sda] Write cache: 
> enabled, read cache: enabled, doesn't support DPO or FUA
> Dec  2 09:58:56 xs2 kernel: [   16.450444]  sda: unknown partition table
> Dec  2 09:58:56 xs2 kernel: [   16.452808] sd 4:0:0:1: [sda] Attached SCSI 
> disk
> Dec  2 09:58:56 xs2 kernel: [   16.991839] EXT4-fs (sda): mounted filesystem 
> with ordered data mode. Opts: (null)
> 
> iscsid isn't running. If I try "service iscsid restart" it doesn't start, 
> though running iscsd by hand works. I also tried the init script "force-
> start" option which does work.
> 
> I think this is because the Fedora 16 init script for iscsid isn't properly 
> detecting that the root is on iscsi. It does
> 
> # FIXME this has a false positive for root on nfs
> root_is_iscsi() {
>     rootopts=$(awk '{ if ($1 !~ /^[ \t]*#/ && $2 == "/") { print $4; }}' 
> /etc/mtab)
>     [[ "$rootopts" =~ "_netdev" ]]
> }
> 
> /etc/mtab is now linked to /proc/mounts, so this doesn't have the _netdev 
> option, even if fstab has this. I think something like "findmnt -m /" is 
> meant to be used instead.

Ah ok.

It should still start becuase we do

    grep -qrs "node.startup = automatic" /var/lib/iscsi/nodes
    if [ $? -eq 0 ] || root_is_iscsi || use_discoveryd ; then

So if you have some records with node.startup = automatic then iscsid
should start. In the iscsiadm output you sent originally you had that right?


> 
> There's some discussion on the fedora-devel list for similar issues in this 
> thread:
> http://lists.fedoraproject.org/pipermail/devel/2011-November/160025.html
> 

Thanks.

-- 
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.

Reply via email to