wissel.net

Usability - Productivity - Business - The web - Singapore & Twins

Moving a NSF from one server to another


I will never FTP a NSF
And no copying with Explorer or SSH either. Moving from one server to another is called replication. If it doesn't work as you expected you have a problem at hand that you need to fix that moment or it will bite you soon enough! (Picture courtesy of Bart Simpson Generator)

Posted by on 26 October 2011 | Comments (d) | categories: IBM Notes Lotus Notes

Comments

  1. posted by Fredrik Norling on Wednesday 26 October 2011 AD:
    I has been many times some administrator thought he be clever and done this. And later removed all documents in the old file Emoticon biggrin.gif because that is a copy Emoticon laugh.gif
  2. posted by Richard Moy on Wednesday 26 October 2011 AD:
    Ran into this problem a few times when customers did not understand Domino.
  3. posted by Gernot Hummer on Wednesday 26 October 2011 AD:
    Similar to my early challenge during designing forms as a fresh Notes dev to keep in mind that hide-whens actually hide stuff when they return @True.

    I had my own fair share of problems with admins copying nsf - files on file system level *grml* Emoticon biggrin.gif
  4. posted by Cristian on Wednesday 26 October 2011 AD:
    I will never FTP a NSF from one server to another....
    WHILE the Domino server is still running Emoticon wink.gif
  5. posted by Stephan H. Wissel on Wednesday 26 October 2011 AD:
    @Christian: NEVER, Domino running or not
  6. posted by Craig Wiseman on Thursday 27 October 2011 AD:
    Unless you're moving the whole server....
  7. posted by Rob Novak on Thursday 27 October 2011 AD:
    Stephan, do you have suggestions for troubleshooting production databases in a staging environment? I run across this all the time with Quickr Domino, and the supported method of moving these nsfs is to just move them over Windows/whatever OS, then register and install them on a target server. Same with place moves.
  8. posted by mark myers on Thursday 27 October 2011 AD:
    if you want to make an admin commit suicide, set the data directories of 2 domino servers to the sync via dropbox, then run like hell.
  9. posted by Keith Brooks on Thursday 27 October 2011 AD:
    Did FTP once many years ago but prefer normal OS copy or some similar way when replication is not possible.
    We used to send out CDs back before better bandwidth.
    Rob, That's how I do it as well for Quickr because you have to.
    Regular nsf from Prod to I have 1 ID which is shared between both domains and is NOT an Admin ID just a generic user.
  10. posted by Stephan H. Wissel on Saturday 29 October 2011 AD:
    @Rob: Take a Notes Admin Client and replicate it over to the other server, then register on the target server.

    @Keith: FTP or OS Level copy are equally bad. What we did when sending CDs (which we had to do a lot): put them on a local machine and as final leg replicate them up to the server

    @Craig: if there is no "another server" involved you can. If you have another one (even temp) replication is the safer option
  11. posted by harkpabst_meliantrop on Friday 31 May 2013 AD:
    Just had to stop by and mention, that "CD-ROM publishing date" is still there in a database's replication settings ... oops, meant to say "Replication Options for this Application" of course ...

    ... and at least as of Notes 8.5.3 it is really still called " CD-ROM publishing date". Emoticon tongue.gif
  12. posted by Don Mottolo on Wednesday 19 February 2014 AD:
    Stephan, Could you explain more of the "why" an OS level or FTP copy is a problem? What are the symptoms/problems you've seen? I can see that at the OS level it would be copying everything bit by bit , including whitespace, deletion stubs, view indexes, corrupted docs/notes, etc. So, a new rep would be cleaner.
  13. posted by Stephan H. Wissel on Wednesday 19 February 2014 AD:
    @Don: you explained it nicely. Only thing to add: each NSF contains an internal master index (actually 2) that keeps track which binary blobs constitute what in the NSF. Think of it as the file allocation table. If both copies get damaged you can throw the NSF away. Creating a new replica creates a new clean index (and doesn't copy view indexes or white space as you pointed out)