[dm-devel] Path updation while controller is rebooted.

[Murthy, Narasimha Doraswamy (STSD)]

Hi,

With latest RHEL 4 DM modules and user space tools, when the Storage
controller is rebooted and if "multipath -ll" is executed, the command
is hung for a while (because of retries) and outputs the following
status:

 

3600508b400102a990000100000710000

[size=4 GB][features="0"][hwhandler="0"]

\_ round-robin 0 [enabled]

  \_ 0:0:0:3 sdc  8:32    [faulty][active]

  \_ 1:0:0:3 sdi  8:128   [faulty][active]

\_ round-robin 0 [enabled]

  \_ 0:0:1:3 sdf  8:80    [faulty][active]

  \_ 1:0:1:3 sdl  8:176   [ready ][failed]

 

IO is not exercised during the above test. I have tested the same by
changing the Qlogic parameters for increasing the timeout/retry values.
The following are the various options used for testing. Each line below
have the Qlogic options which are used for building initrd images.

 

options qla2xxx ql2xretrycount=40 

options qla2xxx ql2xretrycount=60

options qla2xxx  qlport_down_retry=10 ql2xretrycount=60

 

There is not much change in the time taken for executing "multipath -ll"
command.  Just that the command is hung for 40-60 sec.

 

If we have longer retry counts and if the controller takes less time
like a minute to reboot, the status should have been updated as the
contoller is back online within that time.  But in the subsequent path
verification daemon (multipathd) the path status gets updated properly. 

 

The problem is when IO is running on these devices, after the controller
reboots, 

 

"dd" reports -  

 

dd: reading `/dev/3600508b400102a990000100000710000': Input/output error


 

This continues till the next path verification is done. Attached is the
log file.

 

Why does the status not updated correctly? How should I be setting the
HBA (Qlogic in my case) for retry counts and command timeouts ?

 

Thanks and regards

-Murthy

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/dm-devel/attachments/20050817/58bedc13/attachment.htm>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: msg.txt
URL: <http://listman.redhat.com/archives/dm-devel/attachments/20050817/58bedc13/attachment.txt>