Disks from different arrays ends up in a single enclosure (VEA) when using the RDAC driver.
Hi all
System info:
Linux systemname 2.6.18-164.6.1.el5 #1 SMP Tue Oct 27 11:28:30 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux
Red Hat Enterprise Linux Server release 5.4 (Tikanga).
VSF 5, MP3, HotFix3, RollingPatch2.
Sun Microsystems, Inc. RDAC Multi-Path Proxy Driver for Linux, linuxrdac-09.03.0C02.0214.
Arrays:
IBM DS4800
Sun 6580
Before:
When we were using the QLogic failover driver (from the qlogic site) and VSF on top we had disks from different vendors end up in different enclosures (VEA) and that made management easy.
Now:
We are now testing the RDAC driver with the QLogic non-failover driver that RedHat ships with their version 5 distribution of Linux. Now disks from different vendors ends up in one enclosure and with all "Device's" numbered from "disk_0" to "disk_nnnn" it's more difficult to tell the disks appart.
"Internal name" can be changed to something more sensible but with tons of disks it's still best to have different enclosures to deal with.
Has any of you seen this before? What can be done to fix this?
With kind regards
/FrankL
Comments
Hello, Never worked with RDAC
Hello,
Never worked with RDAC driver but I need couple of inputs from you to understand this better...
a) Initially when RDAC driver was not loaded, how was disk names appearing ? If I understand correctly, IBMxxxx for DS series array & SUNxxxx for Sun array, is that correct ?
If above is correct, what I am guessing is since disk names are appearing as disk_xxx, either the ASL is having issue in claiming the devices after RDAC driver installation.This can be tracked by collecting two different outputs of /etc/vx/diag.d/vxasldebug (one before RDAC driver & one after RDAC driver)...
Just an initial thought...
Gaurav
PS: If you are happy with the answer provided, please mark the post as solution. You can do so by clicking link "Mark as Solution" below the answer provided.
Hi Gaurav Thank you for your
Hi Gaurav
Thank you for your reply.
To answer your question (a) i have taken a couple of screenshots (.jpg) and put them up here: http://www.limstrand.net/rdac/
At the same site there is 2 log files, from each run of 'vxasldebug'.
If you 'grep' for "libvxlsiall.so" you will see that the devices are CLAIMED when not using the RDAC driver, and UNCLAIMED when using the RDAC driver. So i guess you are right, the ASL is having an issue in claiming the devices when using the RDAC driver.
Is this support-case material or can i debug this myself? Any suggestions?
/FrankL
Any particular reason why you
Any particular reason why you are using RDAC and not simply Storage Foundation DMP?
The HCL only lists RDAC as a requirement on AIX.
Also have a look at the Hardware TechNote - it lists requirements for DMP functionality.
Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows.
Handy NBU links
Hi Marianne Thank you for
Hi Marianne
Thank you for your answer.
Very long boot times is the problem we experience when we use Storage Foundation DMP on top of QLogic Non-failover driver. If there is a sufficient number of disks we talk about hours to boot a server. Once the server is up everything is fine, preferred paths is followed and so on.
When using the Qlogic failover-driver we have to manually distribute active paths from the HBA driver-side (using SanSurfer - a Qlogic-utility for managing driver behavior) matching active path on our disk systems (DS4800 (IBM) and ST6580 (SUN)). This gives us some headache when thinking about administrating a lot of lun's from different disk enclosures.
The RDAC-driver do this automatically each time it's loading (it is embedded in the ramdisk image linux uses when booting), and we just need to manage active paths from the disk enclosure-side.
It might be an ASL issue, as the devices are UNCLAIMED when using the RDAC driver and CLAIMED otherwise.
/FrankL
Would you like to reply?
Login or Register to post your comment.