I have seen a similar situation in the past. It is undoubtedly not the same thing you are experiencing, but it might shed some light. It will take some explaining. I apologize for the length of this..
We were attempting to use EMC's TimeFinder product. Timefinder allows an additional set of mirror volumes to be associated with logical volumes in an EMC Symmetrix system. These mirrors, called "BCVs" or "Business Continuance Volumes", are said to be "established" when they are actively being kept in sync with their associated master volumes. At a time of your choosing you can "split" them, which means that the mirroring stops, and the BCV volumes represent a snapshot of the master volumes at a moment in time. This is analogous to splitting mirrored pairs on Mirrored Disk/iX. When split, those BCV volumes can be mounted by the same or another system. When they are established, the BCVs are invisible.
We were using TimeFinder to provide a copy of a production volume set on a test system. We could split the BCVs, make them available for testing, reporting, etc. all day, then re-establish them that night. The Symmetrix keeps a log of all modified tracks, and re-syncs the BCVs with the masters very quickly. As I said, when you establish the BCV's they become invisible to the second system. It is as if they have been unplugged. When we were ready to do an establish operation, we would VSCLOSE the BCV volume set on the test machine, then perform the establish. When we wanted to use the BCVs again, we would wait for a time when the system was quiet, split the volumes, and VSOPEN them on the test machine.
The problem came when we did the split. It takes several seconds for the Symmetrix machine to split all the volumes, and if splitting them as a group, they become available in an unpredictable order. The problem is that the test system would not wait for the operator to enter a VSOPEN command to mount the volume set. As soon as it saw the master volume come online, it would automatically begin opening the volume set. Unfortunately, it would try to mount some of the volumes before they were online, and those volumes would perpetually get stuck in the "Busy" state that you are seeing. A system reboot was the only cure. Once we understood the problem, we prevented it by splitting all the member volumes first, then splitting the master after waiting a few seconds to make sure all were available for mounting.
I suspect your problem relates to a LUN being unresponsive for whatever reason while the system was trying to open the volume set. I hope this sheds some light on your problem. Good luck.
-----Original Message-----
From: David Knispel [mailto:[log in to unmask]]
Sent: Friday, September 26, 2003 10:18 AM
To: [log in to unmask]
Subject: HP3000-XP512 Problem
Help!
Here's the problem. My HPe3000 MPEiX 7.5 system is locked up and has been
since about 8am this morning. I can still log onto the system using
;PARM=-1 because what appears to be locked is my XP and not my internal boot
drive. When I do a DSTAT all one of the XP presented LUN's is showing as
below...
...
234-OPEN-E MEMBER MEMBER33 XP01_SET-0
235-OPEN-E MEMBER MEMBER34 XP01_SET-0
236- BUSY MEMBER MEMBER35 XP01_SET-0
237-OPEN-E MEMBER MEMBER36 XP01_SET-0
238-OPEN-E MEMBER MEMBER37 XP01_SET-0
...
We've replaced the fiber cable and it still acts the same. I've not
rebooted the 3000 yet. Might that help? Has anyone seen this before? HP
has been here since this morning and they are puzzled. I thought I'd try
asking you all to see if you had any ideas...
David Knispel
[log in to unmask]
Phone: 513-248-5029
Fax: 513-248-2672
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|
