UNSOLVED

srichev

updated

18 years ago

S

srichev

130 Posts

0

1386

January 29th, 2009 13:00

Clones on R2 side

As per Primus solutions,In order to perform a clone on R2-side one full copy is needed before activating in a consistent mode for SRDF/A.Can this be incremental after this? or does this needs to be full copy every time?
I am listing the commands also..let me know,if they are(not)going to work for R2 side.

Symclone -g create -precopy -diff -rdf -tgt (For the initial establishment,or is the -copy -diff going to work also?)
symclone -g activate -consistent -rdf -tgt

let me say,I need to sync the clone day after with changed data,Can i issue,
symclone -g establish -consistent -rdf -tgt

Also,I am planning to use the -both_sides with the establish but it seems its giving me an error (not a valid option)

symclone -g establish -consistent -both_sides

Your suggestions..

null
  • xe2sdc

    6 Operator

    •

    2831 Posts

    667

    0

    Posted January 30th, 2009 02:00

    When making clones with SRDF/A on the R2 side, you need to use PRECOPY.
    And you need precopy even when you recreate the clones. However you can clone incrementally (after the first full copy).

    Since you need precopy, -copy won't work :-)
    Before activating the clones you need to wait a full precopy cycle.

    symclone -g -tgt -rdf -precopy -cycled verify

    (or check symclone query output)

    When precopy did its dirty work, you can activate consistently your clone :-)

    In case you want to "refresh" the clones, you need to recreate them (again with precopy option) wait for a full precopy cycle (same command as above) and finally activate consistently the clones :-)

    Are you using clones on both sides (R1/R2) ??

    Message was edited by:
    Stefano Del Corno

    add recreate directions :D
  • srichev

    130 Posts

    667

    0

    Posted January 30th, 2009 05:00

    yes using clone on both directions.My concern was,on R2 side,Is the full pre-copy required everytime. I was checking this option of

    symdg -g recreate -precopy

    If the first create has -precopy -diff,remaining going to be just the changes instead of full copy.Is the above going to work?

    What is the option -both_sides?Is this going to allow me to activate clones at the same time on R1 and R2 sides?
  • xe2sdc

    6 Operator

    •

    2831 Posts

    667

    0

    Posted January 30th, 2009 05:00

    Unfortunatly I don't have experience with -both_sides option ... If possible I'll dig a little next week with this feature ..

    Note that when you create session with -differential, the recrate will obviously be incremental. The pre-copy process doesn't force a full copy. Enabling pre-copy means you start copying tracks right after CREATING/RECREATING the pairs. Thus when you recreate sessions, you'll start copying changed tracks.

    Please note that the procedure I gave you is what my beloved customer uses twice every day on 20 servers ;-) .. Thus I can tell you it works :-)
  • xe2sdc

    6 Operator

    •

    2831 Posts

    669

    0

    Posted January 30th, 2009 06:00

    you mean customer was doing full copy
    everytime?


    No I mean customer is doing incremental copies with precopy enabled.

    Recreating (with precopy) a differential clone will copy only changed data (no full copy). :-)
  • srichev

    130 Posts

    667

    0

    Posted January 30th, 2009 06:00

    you mean customer was doing full copy everytime?Wondering hwo long is it taking(i need to make a clone for roughly 10TB daily once).
    If you can check the both sides option that would be great.
  • srichev

    130 Posts

    669

    0

    Posted January 30th, 2009 07:00

    Thanks Stefano :)
  • Allen Ward

    6 Operator

    •

    2124 Posts

    669

    0

    Posted February 2nd, 2009 08:00

    We are doing the same thing, and to answer the one question that may have got lost (or glossed over) in the rest of the thread, if you are doing clones of an R2 the -precopy is definitely required. You need to have a script issue the precopy and scan the output of queries until the precopy completes one cycle before issuing the activate.

    As Stefano pointed out, the precopy cycle after the initial activate is only an incremental update even though we talk about a "full cycle". The precopies after the first activate tend to be fairly fast... but that will depend a lot on your rate of change.
  • xe2sdc

    6 Operator

    •

    2831 Posts

    669

    0

    Posted February 2nd, 2009 11:00

    Allen we tend to use symclone with appropriate options to check if precopy did its dirty job ;-)

    symclone -sid xxx -file pairfile -precopy -cycled verify

    The command will return $? = 0 if all devices in pairfile (or if you prefer in your DG) already completed the precopy. Any other return code means at least a device is still precopying (poor wording but hope it's clear) :D
  • Allen Ward

    6 Operator

    •

    2124 Posts

    669

    0

    Posted February 2nd, 2009 12:00

    Oddly enough I suggested that the support team managing the clone process use "verify" to monitor progress, but they preferred to write a script that checked the output of a "query". I'm not sure if there was additional info they wanted to capture or what, but I had completely forgotten about using verify since my guys didn't want to use it.

    It is actually another team that manages the clone process since it is part of a business process that has to tie into quiescing the database on the R1 side of the SRDF/A link. Storage is only responsible for making sure SRDF/A stays up.
  • xe2sdc

    6 Operator

    •

    2831 Posts

    136

    0

    Posted February 2nd, 2009 22:00

    It is actually another team that manages the clone
    process since it is part of a business process that
    has to tie into quiescing the database on the R1 side
    of the SRDF/A link. Storage is only responsible for
    making sure SRDF/A stays up.


    Thus you have to carefully check that cycles switches every 30 seconds .. and you have to check that frames flow to the R2 side .. and that the R2 side applies them consistently ?? Not that bad ;-)