Bug 212102 - linux-iscsi ISCSI_MAX_CMDS_PER_LUN should be larger to allow for perf tuning
Summary: linux-iscsi ISCSI_MAX_CMDS_PER_LUN should be larger to allow for perf tuning
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Enterprise Linux 5
Classification: Red Hat
Component: kernel
Version: 5.0
Hardware: All
OS: Linux
medium
medium
Target Milestone: ---
: ---
Assignee: Mike Christie
QA Contact: Brian Brock
URL:
Whiteboard:
Depends On: 181787
Blocks:
TreeView+ depends on / blocked
 
Reported: 2006-10-25 03:14 UTC by Dave Wysochanski
Modified: 2009-11-10 18:34 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: Enhancement
Doc Text:
Clone Of:
Environment:
Last Closed: 2009-11-10 18:34:34 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)

Comment 1 Dave Wysochanski 2006-10-25 03:21:18 UTC
Might be too late for RHEL5 but perhaps changing the max #define isn't out of
the question. (Sanjay Gulabani at netapp asked for this -
Sanjay.Gulabani)

Here's the thread earlier this year with a patch that never got in upstream:
http://tinyurl.com/wynfp

Maybe we can resurrect something like that for upstream and long-term.


Comment 2 Dave Wysochanski 2006-10-25 13:18:43 UTC
Ok I looked at the code in more detail and remember that it's probably not so
easy to change the #define because of the structures we allocate based on these
max values.

I was going to reassign this to myself but it doesn't look like I can change the
assignment field.

Comment 4 Ernie Petrides 2007-09-05 22:00:38 UTC
Reassigning to DaveW based on comment #2.

Please fix the bug summary to contain something that can be found in
the RHEL5 kernel source tree.  ISCSI_MAX_CMDS_PER_LUN does not exist,
and the place in RHEL4 where its use was converted to iscsi_cmds_per_lun,
which was in the function iscsi_slave_configure(), doesn't exist in RHEL5.

Comment 5 Mike Christie 2007-09-05 22:12:10 UTC
Dave I can take this one if you want. It is all upstream in the open-iscsi git
tree so it just requires a backport. I did not port this part in my 5.1 update
because I did not have time to test the KABI and ABI workarounds that are needed
to port upstream to RHEL.

Comment 6 Dave Wysochanski 2007-09-05 22:21:37 UTC
Ok thanks - I am sure it would take me much longer to get it right.

Comment 7 Mike Christie 2009-11-10 18:34:34 UTC
This was fixed in RHEL 5.2 or RHEL 5.3. Closing bug.


Note You need to log in before you can comment on or make changes to this bug.