recover from many tapes at the same time

[frankr]

Hi

im having problems recovering from many tapes in parallel. Earlier it worked , but when i upgraded to networker 6.x it stopped working - have anyone a solution to this - i already have the file striped_recover in my debug lib. the problem occurs when i have to recover from more tapes than i have tapedevices.

Frank

 

[wc310]

Call legato support, there is a fix for this.

 

[frankr]

hi

i already did that, and the fix only works if the data to restore is on less tapes than i have tapedevices.

Frank

 

[RMinNJ]

Frank,

What is the fix or how do you get a 6.x server to mount and
read tapes in parallel during a recover? My restores
mount each tape one at time ..ie if I have 4 tapes for a
recover it mounts and reads the first, when its done it
mounts and reads the second...etc, even though I have 10
drives. I'd be happen for any parallelism.

Robert

 

[frankr]

thats easy

in Your debug library / folder or whatever you want to call it.

NT it is normally c:\nsr\debug , on Unix it is normally /nsr/debug

in this location you make an empty file (touch in unix)

file name should be , striped_recover

no extensions , thats it.

Frank

 

[frankr]

I almost forgot.

There is a slight problem withthis fix. ALL the volumes needed for the recover has to be mounted at the same time!!

meaning if you have 10 drives and the data you want to recover is on 11 tapes , it wont work - then you have to split the recover into smaller parts - but it is still faster than the other way. My recover speed is about 400 GB in 3 hours


Frank

 

[RMinNJ]

frank,

Thanks for the procedure..sounds easy. If you have more
tapes than drives for the restore are you saying the restore
fails or does it revert back to one tape at time?

I suppose we could turn it off for large multiple tape
restores although that's precisely the case where we would
want it on..

Robert

 

[frankr]

hi Robert

remember to restart networker to make it work

it just dont start - it waits for the tapes, that never comes. it cant revert to normal. I normally divide my recovers into smaller parts that fits into the 8 drives i have. normally my SAP incl. Unix operating files and all the other stuff we have is on 16 to 20 tapes , so i would recover - say SAPDATA1 to 8 in one and when it is finished i would take the unix part. as a matter of fact it is very easy, when in the recover program - you add files or directories , after each add , you can type "volumes" to see how many volumes are needed for the recover , just add until you are over the limit and then remove one directory and go.

Frank

 

[RMinNJ]

Frankr,wc310,

I spoke to Legato and they said this is a debugging technique and should not be used in a production
environment.

I did gather the client needs extra disk space to rearrange
savestreams that come in out of order.

wc310 is there really a fix out for frankr's problem? I
can only assume its a unsupported feature; fast, but not something I'll get support on if a restore fails.

 

[frankr]

Yes it is true that it is a debugging technique , but i have used it for a year now, Legato told me to do it this way. Earlier, in version 5.2 the feture was there - full blown with no restrictions, but they had some problems with another part of the software so they disabled this "multi device recover" function. Legato tells me that it will be back in version 7, release date Q4 2002