Hi Mike,

Mike Christie wrote:
> Rudy Gevaert wrote:
>> Hello,
>> So it seems that running 2.6.18 (stock Debian kernel) with open-iscsi 
>> 2.0.730-1etch1 doesn't give me any problems regarding the start and 
>> stopping.  (Although one iscsid process is still running when I stop it 
>> with the init script.)
> That is a bug in iscsid not handling the kill signals right. Other 
> scripts do a killall -9 iscsid type of thing but debian does not. Now 
> iscsid handles the normal stop signals so there is no need to do killall -9.

I can say that it doesn't affect me if that one process is still 
running.  Or I haven't experienced any side effects.

>> Doing
>>   echo - - - > /sys/class/scsi_host/hostX/scan
>> effectively rescans my target, so the need to upgrade to a later version 
>>   is not necessary (for me).
>> However using a later open-iscsi (2.0.865-1) has some problems on 2.6.18 
>> kernel.  It isn't really stopping when I stop it.  The devices are still 
>>    available.
> I do not think 2.0.865 works with the 2.6.18 upstream kernel. The 
> interface was not stable until 2.6.21, so before that you had to upgrade 
> tools with the kernel or the distro kernel had to be updated.

Correct, it works up to a certain hight.

>> Debian bug http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=467390 has 
>> some notes about the newer version of open-iscsi not working with 
>> kernels below 2.6.25.
> I am not sure if that is worded right. The scsi userspace interface we 
> used was broken and that broke the open-iscsi tools. The newest version 
> of open-iscsi tools contain a fix, but they require a kernel fix that 
> is in 2.6.25.
> Alternatively, you can use the kernel modules and tools in the 
> open-iscsi.org release because they have all the updated fixes.
> Also, I am not sure if this 
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=467390 affects you. 
> With netapp targets (just guessing you have a netapp target because of 
> the naming of your rule below) I think you can control the write back 
> cache settings.  If it is disabled like here:

Luckily it doesn't affect me.

> Apr  1 12:04:07 meanminna kernel: sd 9:0:0:8: [sdo] Write cache: 
> disabled, read cache: enabled, doesn't support DPO or FUA
> then you will not hit the problem.
>> 2.6.24 is in Debian unstable, without xen for amd64 and I need xen 
>> support, so I'm rather stuck with 2.6.18.
>> Stopping open-iscsi also deletes the correct /dev/disk/by-path links. 
>> But I'm still having problems with a udev rule I wrote myself:

<udev script>

>> Are there any guidelines for integrating udev with open-iscsi?
> No. There are some threads on the list from users doing their custom 
> scripts, but normally we probably prefer to user the distros default 
> rules. If you have the most uptodate ones they should normally be fine 
> for what you want to do. If they are not then we should talk about why not.

The default rules are ok for iscsi.  I have know figured out the 
following.  If one is depending on /dev/device to be available then one 
must use the RUN keyword.  Executing programs with PROGRAM or 
IMPORT{program} doesn't guarantee that /dev/device is available at the 
moment the program is run.  Using RUN guarantees that.

The downside of using RUN is that one has to manage everything by hand. 
  E.g. the adding/removal of symlinks.

Thank you for helping me getting to the bottom of this!

-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
Rudy Gevaert          [EMAIL PROTECTED]          tel:+32 9 264 4734
Directie ICT, afd. Infrastructuur ICT Department, Infrastructure office
Groep Systemen                    Systems group
Universiteit Gent                 Ghent University
Krijgslaan 281, gebouw S9, 9000 Gent, Belgie               www.UGent.be
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --

You received this message because you are subscribed to the Google Groups 
"open-iscsi" group.
To post to this group, send email to open-iscsi@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/open-iscsi

Reply via email to