This is fixed in iscsi-initiator-utils-6.2.0.872-14.el6. You can download it here
http://people.redhat.com/mchristi/iscsi/rhel6.1/iscsi-initiator-utils/

See the iscsiadm man page's EXIT STATUS section for info on the error codes.

Mike,
Do you have any code for testing all of these error?
If not, I will try to create one.

No code. Sorry.

If you can create a script where it runs iscsiadm then checks the return code for operations, I will integrate it into open-iscsi upstream. Then I can also run it when merging patches and before even making a rpm for you to test. So it should help you out.

(In reply to comment #6)
> If you can create a script where it runs iscsiadm then checks the return code
> for operations, I will integrate it into open-iscsi upstream. Then I can also
> run it when merging patches and before even making a rpm for you to test. So it
> should help you out.

It will take some for finishing this tool.

When I am reviewing the code and try to hit the error code, I found a typo in the code of 6.2.0.872:
usr/iscsi_err.c 

47c47
<       /* 22 */ "sysfs lookup failure",
---
>       /* 23 */ "sysfs lookup failure",

It's just a comment and you might change it for next respin.

Q1:
For ISCSI_ERR_SESS_NOT_FOUND error, any hit for hit it?
I tryed this command: iscsiadm -m session -r 99
But got $? 22 ISCSI_ERR_SYSFS_LOOKUP instead of 4 ISCSI_ERR_SESS_NOT_FOUND

Q2: 
For ISCSI_ERR_LOGIN error, I try to use iptables to DROP package to 3260 port.
'iscsiadm -m node -l' will goes into a infinite sleep and waiting for iscsid to return.
If I Ctrl-C, iscsid still there and waiting for package which cause iscsiadm login failed with 15 - session exists.
Do you think this is bug, if so I will file a new one for it.

(In reply to comment #7)
> <       /* 22 */ "sysfs lookup failure",
> ---
> >       /* 23 */ "sysfs lookup failure",
> 
> It's just a comment and you might change it for next respin.
> 

Thanks. Fixed upstream. Will bring in fix in next respin.

> Q1:
> For ISCSI_ERR_SESS_NOT_FOUND error, any hit for hit it?
> I tryed this command: iscsiadm -m session -r 99
> But got $? 22 ISCSI_ERR_SYSFS_LOOKUP instead of 4 ISCSI_ERR_SESS_NOT_FOUND

Ah yeah, ISCSI_ERR_SYSFS_LOOKUP makes sense from a devel perspective. The sysfs code could not find the session so that is why you get the error. From a user perspective ISCSI_ERR_SESS_NOT_FOUND makes a lot more sense. Will fix that on the next respin.

I do not think there is a easy way to hit ISCSI_ERR_SESS_NOT_FOUND(2). It only occurs during races where if you were to run iscsiadm twice and both instances where going to logout the same session. If you can get the timing right iscsid would return ISCSI_ERR_SESS_NOT_FOUND. If once iscsiadm instance completely logs out a session and cleans up sysfs before the other iscsiadm could read the sysfs info, then the other iscsiadm instance would get ISCSI_ERR_SYSFS_LOOKUP.



> 
> Q2: 
> For ISCSI_ERR_LOGIN error, I try to use iptables to DROP package to 3260 port.
> 'iscsiadm -m node -l' will goes into a infinite sleep and waiting for iscsid to
> return.

Is it infinite or just a long time? The default wait time is a couple minutes.

If you set node.conn[0].timeo.login_timeout to 10 and node.session.initial_login_retry_max to 0, then it should be really quick.


If it is not real quick with those settings. It is a bug. Make me a bz and I will fix it. Could you also run iscsid by hand if it is a bug. Do

iscsid -d 7 -f &


and send me all the log output that gets printed.



> If I Ctrl-C, iscsid still there and waiting for package which cause iscsiadm
> login failed with 15 - session exists.

Do you mean when you do Ctrl-C then when iscsiadm returns it returns with 15? I tried it here and got 1, which is expected (1 is just a generic error).

Or do you mean you did Ctrl-C then when you reran iscsiadm it returned 15? This is sort of expected. Ctrl-C will kill iscsiadm, but iscsid is still executing the operation, so it sort of still exists in iscsid. If this is what you are hitting I will make a bugzilla. I will try to fix it. It is not a simple change is the reason it has not got fixed yet.

(In reply to comment #8)
Is it infinite or just a long time? The default wait time is a couple minutes.

Yes. By default, it's 120s for timeout. It quit as error 8. My fault.

As I have tested most of the return code. I would like to VERIFY this bug with rhts tag.

Storage-qe are out of capacity of RHEL6.1 and RHEL 5.7 testing, for this exit status script, we will try out best to finished it.

    Technical note added. If any revisions are required, please edit the "Technical Notes" field
    accordingly. All revisions will be proofread by the Engineering Content Services team.
    
    New Contents:
When iscsiadm failed or exited incorrectly, it did not output useful error codes. Meaningful error codes now exist for these situations, and are described further in the iscsiadm man page.

An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHBA-2011-0733.html