<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">On Monday 29 October 2012 07:32 PM,
Dennis Leeuw wrote:<br>
</div>
<blockquote cite="mid:508E8C64.4060303@umcutrecht.nl" type="cite">We
can make this a long discussion, but I rather not. I do not have
the knowledge to tell you what might or might not happen and why.
I can only tell you what went wrong and restarting the open-iscsi
service, without a block device being present created a problem
with a ro / filesystem. But actually that is not the important.
What is important is that if no block device is present, the
script does not end gracefully, with the provided change it does.
<br>
<br>
</blockquote>
<br>
What you see out there is not something very simple. And given my
limited time I spend on Debian, I will investigate in detail only if
I am convinced it to be a bug.<br>
<br>
That said, support for sw iSCSI block devices is more complicated
than what it looks. On top of that, adding the Device Mapper layer
further adds a layer of complications. In the current approach
available, you at least have an option to have your entire block
device auto-mounted/activated during boot.<br>
<br>
Remember, iscsi is a daemon. If the daemon is stopped, all your
block devices could die. That's why the stop script is built with
some assumptions. More below....<br>
<br>
<blockquote cite="mid:508E8C64.4060303@umcutrecht.nl" type="cite">running
without my change:
<br>
# /etc/init.d/open-iscsi restart
<br>
Unmounting iscsi-backed filesystems: Unmounting all devices marked
_netdevDeactivating iSCSI volume groups: md3200_1 Volume group
"md3200_1" not found
<br>
Cannot deactivate Volume Group md3200_1 ... (warning).
<br>
.</blockquote>
<br>
The assumption being that if you have DM devices on top of your
iSCSI devices, we expect that you (the user) will define them in the
config file. That is what you did.<br>
<br>
You instructed it to take care of a non-existent VG, which it did,
failed and complained.<br>
<br>
This is the current design. It might not be the most elegant, but it
works better than many others I have seen.<br>
<blockquote cite="mid:508E8C64.4060303@umcutrecht.nl" type="cite">
<br>
umount: /: device is busy.
<br>
(In some cases useful info about processes that use
<br>
the device is found by lsof(8) or fuser(1))
<br>
Could not unmount / ... (warning).
<br>
umount: /var: device is busy.
<br>
(In some cases useful info about processes that use
<br>
the device is found by lsof(8) or fuser(1))
<br>
Could not unmount /var ... (warning).
<br>
failed!
<br>
invoke-rc.d: initscript umountiscsi.sh, action "stop" failed.
<br>
Couldn't unmount all iSCSI devices. Cannot stop iSCSI service ...
failed!
<br>
<br>
</blockquote>
<br>
I can't see a reason why / and /var got pulled in... Can you run the
same init script in debug mode?<br>
Just add "set -vx" to the beginning of the umountiscsi.sh script so
that we can see what is going on.<br>
<br>
<blockquote cite="mid:508E8C64.4060303@umcutrecht.nl" type="cite">Which
is probably not very user friendly. It doesn't matter if my system
get Read-Only or not.
<br>
<br>
With the following lines added:
<br>
if [ "${BLOCK_FILE}" = '<i class="moz-txt-slash"><span
class="moz-txt-tag">/</span>sys/devices/platform/host24/session20/target*/*:*/block<span
class="moz-txt-tag">/</span></i>*' ]; then
<br>
log_warning_msg "No block devices found for this host,
nothing to do"
<br>
return 1
<br>
fi
<br>
<br>
It ends more gracefully:
<br>
# /etc/init.d/open-iscsi restart
<br>
Unmounting iscsi-backed filesystems: Unmounting all devices marked
_netdevDeactivating iSCSI volume groups: md3200_1 Volume group
"md3200_1" not found
<br>
Cannot deactivate Volume Group md3200_1 ... (warning).
<br>
.
<br>
No block devices found for this host, nothing to do ... (warning).
<br>
invoke-rc.d: initscript umountiscsi.sh, action "stop" failed.
<br>
Couldn't unmount all iSCSI devices. Cannot stop iSCSI service ...
failed!
<br>
<br>
open-iscsid is still running and a restart didn't happen, but that
is another issue.
</blockquote>
<br>
<br>
<pre class="moz-signature" cols="72">--
Ritesh Raj Sarraf
RESEARCHUT - <a class="moz-txt-link-freetext" href="http://www.researchut.com">http://www.researchut.com</a>
"Necessity is the mother of invention."
</pre>
</body>
</html>