[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference cookie::raid_software_for_openvms

Title:	RAID Software for OpenVMS
Notice:	READ IMPORTANT NOTE IN 3.15, V2.4 SSB Kit in 3.176
Moderator:	COOKIE::FROEHLIN

Created:	Fri Dec 03 1993
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	341
Total number of notes:	1378

326.0. "DPA take longer to show up after CLUSIO01 is installed" by TIMABS::FREPPEL (Mosquito ergo summm...) Tue Mar 11 1997 17:33

    Hi all,

    in the customer's cluster, all nodes bind and mount a bunch of DPA 
    devices during system startup.

    As soon as the DPAxxx's are bound, the subsequent booting nodes get the
    following message:
    %RAID-E-ALREADYBOUND, RAID array is already bound
    which is normal.

    The strange thing is that sometimes the subsequent MOUNT command
    MOUNT/SYSTEM DPAxxx: label logical
    brings back the following error message:

    %MOUNT-F-NOSUCHDEV, no such device available

    this is also the case when we wait 10seconds between the BIND and the
    MOUNT commmand. This happens to different DPA devices.

    10 seconds seem fairly long to me for the device to appear, and as of
    now  this error did not occur.
    Do we have to explicitely check (a la f$getdvi("dpaxxx:","exists"))?
    Other suggestions?

    This is OpenVMS Alpha V6.2 and V6.2-1H2 and OpenVMS VAX V6.2
    VAXCLUSIO01_062
    ALPCLUSIO01_062
    RAID Software for OpenVMS V2.3

    Thanks.
    Raymond.

T.R	Title	User	Personal Name	Date	Lines
326.1		COOKIE::FROEHLIN	Let's RAID the Internet!	`Wed Mar 12 1997 14:14`	33
	Hi Raymond! > %RAID-E-ALREADYBOUND, RAID array is already bound Doesn't mean the DPA device is already configured. The newly started RAID server has received information about an existing array from another node. So it knowd about the array. But it still has to mount the member devices and tell the local driver about the array and the virtual devices. This might take a while. Depending on the number of arrays it can take more than 10 seconds to have a peticular DPA device showing. A F$GETDVI("DPAXXX:","EXISTS") and a wait loop is a good solution. > MOUNT/SYSTEM DPAxxx: label logical > %MOUNT-F-NOSUCHDEV, no such device available Means the local server has not reached the point where it could mount mount the member devices and tell the driver about this array. > 10 seconds seem fairly long to me for the device to appear, and as of The server is working on one array at a time. If the procedure happens to wait on a DPA device from the last array then yeah, it'll take longer. Roughly it takes 3-4 seconds to mount one member device. I don't see and impact from the VMS patches installed. As a standard system management action I would put these activities into a batch job, maybe several batch jobs, in order to keep the main both thread going. Guenther
326.2		TIMABS::FREPPEL	Mosquito ergo summm...	`Wed Mar 12 1997 14:47`	20
	Thanks, Guenther, for the explanation. I'll give the f$getdvi(mumble)-wait-loop a try. And if the loop doesn't end within reasonable amount of time I will post another reply here... > I don't see and impact from the VMS patches installed. I'm still wondering what could have caused the sudden change in behaviour (on two separate nodes with two different DPA devices). > As a standard system management action I would put these activities > into a batch job, maybe several batch jobs, in order to keep the main > both thread going. The startup is already organized in several threads. Thanks again, Raymond. PS: Over all the customer is very happy with your product, congratulations to you and your team!