So here's what I came up with. Probably no surprises, except maybe that
the NFS backend machine needs some care. This is because the OpenGrok
backend runs on it, and the OpenGrok backend does pulls.
Rich, do you have any thoughts on the testbed setup? Is that something
you want to tackle? Joint project between you and me...?
Also, if anyone has comments on the things marked as issues, please
speak up.
thanks,
mike
--
* scm1 Solaris upgrade
upgrade to snv_79 or later (to get Mercurial 0.9.5)
- who: tonic-ops
- when: Dec 2007? Jan 2008?
* testbed setup
set up dummy gate (whatever is sufficient for testing gate hooks)
- who: ???
- when: after scm1 Solaris upgrade
* update and test gate hooks
update gate hooks (if needed), test, commit, generate OSOL packages
- who: ???
- when: after testbed setup
- issue: should we move the gate hooks out of scm-host-repo?
- issue: is current OSOL package organization/granularity correct?
- issue: the current upgrade instructions are pretty heavy-weight
(assumes pkgrm and pkgadd of more than just the hooks); what's the
minimum set of steps if we're just updating Python extensions?
** tests
- push w/ 1 changeset
- push w/ multiple changesets
- anon pull
- pull as authenticated user
** followup
- docs changes as needed
- scm-host-repo/doc/d-scm-installation.txt
* 1st app server
upgrade Solaris to same as scm1, install new OSOL package(s),
sanity test.
- who: ???
- when: after new OSOL packages are ready
* 2nd app server
same as "1st app server", but wait until 1st app server has been
completed
* NFS server
upgrade Solaris, install new OSOL package(s), test.
- who: ???
- when: after new OSOL scm packages are ready
- issue: do we do a rolling upgrade or schedule downtime?
** tests
- sanity test of scm operations
- OpenGrok
