OPT Roll back instructions

Dave Harland

Let's say version N of model-to-script test program (m2s) is in production and we want to retreat to version N-1.

  • Lucky Scenario: Version N-1 can successfully fetch SBs from Version N Database
    • This is less strict than "database structure has not changed from N-1 to N" because sometimes the change in structure is benign. This would be the case, for example, where we've added new columns and/or tables but have done no violence to columns and tables that were present in N-1.
    • In this case one would check out from subversion tag N-1 of m2sTest, the svn root of which is https://svn.aoc.nrao.edu/repos/SSS/m2sTest/. One would then run "mvn clean install" to build the jar file. One would next copy the jar file and m2sStable.jnlp to /home/webtest.aoc.nrao.edu/content/dharland/m2s. (Remember m2s was intended as a test pgm so it has a pretty unofficial home.)
    • Shortcut: So far i have been keeping all old jar and jnlp files in an "old/N" subdirectory of the above webtest directory. Merely protecting the current files (ie, renaming them) in the m2s root and copying over the N-1 files to the root would do the trick much more quickly.
    • Any of the SSS pgmrs (Witz, Truitt, Lyons, Harland) s/b able to do this.

  • Unfortunate Scenario: Version N-1 cannot deal w/ Version N Database
    • In this situation a new hybrid version {N-1, N} would need to be built. One would be trying to build m2s so that it had the script-creation-logic of version N-1 but was in synch w/ version N of the database. This might be easy or it might be so difficult as to not warrant doing. "Difficult" means that version N-1 of the model library and version N of database library cannot be compiled together. If that's the case, one would probably create version N+1 of m2s, changing the logic in whatever way warranted the attempt to retreat.
    • Probably any of the SSS pgmrs could attempt the N-1, N hybrid build. If it came to a coding effort, it s/b Truitt or Harland.

If we felt we had to roll back the m2s test pgm, then we probably would want to also roll back the OST. I think i first need to clarify where the logic of turning SBs into scripts lies.

The logic that creates a script from an SB is part of the sss-model jar file (a compiled library). When we compile the OPT, it includes this jar file. When we compile the OST, it includes this jar file. When we compile m2s, it includes this jar file.

If all we do is roll back m2s to N-1, the OST (and less importantly, the OPT) still has the logic of version N. Rolling back m2s has no impact on the OST or OPT.

-- JosephMcMullin - 10 Aug 2010
Topic revision: r1 - 2010-08-10, jmcmulli
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback