5 - Keep my sanity when upgrading 1000 installs
6 - Replace gaierror with a more descriptive name (this is a DNS error)
8 - Make the rest of the world use Wizard
9 - Make parallel-find.pl use `sudo -u username git describe --tags`
10 to determine version. Make parallel-find.pl have this have greater
11 precedence. This also means, however, that we get
12 full mediawiki-1.2.3-2-abcdef names (Have patch, pending testing and commit)
13 - Make deployed installer use 'wizard install' /or/ do a migration
14 after doing a normal install (the latter makes it easier
17 - Pre-emptively check if daemon/scripts-security-upd
18 is not on scripts-security-upd list (/mit/moira/bin/blanche)
20 - Redo Wordpress conversion, with an eye for automating everything
21 possible (such as downloading the tarball and unpacking)
24 - Genericize callAsUser and drop_priviledges in shell
25 - Summary script should be more machine friendly, and should not
26 output summary charts when I increase specificity
27 - Summary script should do something intelligent when distinguishing
28 between old-style and new-style installs
29 - Report code in wizard/command/__init__.py is ugly as sin. Also,
30 the Report object should operate at a higher level of abstraction
31 so we don't have to manually increment fails. (in fact, that should
32 probably be called something different). The by-percent errors should
34 - Indents in upgrade.py are getting pretty ridiculous; more breaking
35 into functions is probably a good idea
36 - Move resolutions in mediawiki.py to a text file? (the parsing overhead
38 - Investigate QuotaParseErrors
39 - If a process is C-ced, it can result in a upgrade that has
40 an updated filesystem but not updated database. Make this more
42 - PHP end of file allows omitted semicolon, can result in parse error
43 if merge resolutions aren't careful.
46 - Make single user mass-migrate work when not logged in as root
47 - Don't use the scripts heuristics unless we're on scripts with the
48 AFS patch. Check with `fs sysname`
49 - Make 'wizard summary' generate nice pretty graphs of installs by date
50 (more histograms, will need to check actual .scripts-version files.)
51 - It should be able to handle installs like Django where there's a component
52 that gets installed in web_scripts and another directory that gets installed
54 - ACLs is a starting point for sending mail to users, but it has
55 several failure modes:
56 - Old maintainers who don't care who are still on the ACL
57 - Private AFS groups that aren't mailing lists and that we
59 A question is whether or not sending mail actually helps us:
60 many users will probably have to come back to us for help; many
61 other users won't care.
63 PULLING OUT CONFIGURATION FILES IN AN AUTOMATED MANNER
65 advancedpoll: Template file to fill out
66 django: Noodles of template files
67 gallery2: Multistage install process
69 mediawiki: One-step install process
70 phpbb: Multistage install process
71 phpical: Template file
74 wordpress: Multistage install process
76 PHILOSOPHY ABOUT LOGGING
78 Logging is most useful when performing a mass run. This
79 includes things such as mass-migration as well as when running
80 summary reports. An interesting property about mass-migration
81 or mass-upgrade, however, is that if they fail, they are
82 idempotent, so an individual case can be debugged simply running
83 the single-install equivalent with --debug on. (This, indeed,
84 may be easier to do than sifting through a logfile).
86 It is a different story when you are running a summary report:
87 you are primarily bound by your AFS cache and how quickly you can
88 iterate through all of the autoinstalls. Checking if a file
89 exists on a cold AFS cache may
90 take several minutes to perform; on a hot cache the same report
91 may take a mere 3 seconds. When you get to more computationally
92 expensive calculations, however, even having a hot AFS cache
93 is not enough to cut down your runtime.
95 There are certain calculations that someone may want to be
96 able to perform on manipulated data. As such, this data should
97 be cached on disk, if the process for extracting this data takes
98 a long time. Also, for usability sake, Wizard should generate
99 the common case reports.
101 Ensuring that machine parseable reports are made, and then making
102 the machinery to reframe this data, increases complexity. Therefore,
103 the recommendation is to assume that if you need to run iteratively,
104 you'll have a hot AFS cache at your fingerprints, and if that's not
105 fast enough, then cache the data.
107 COMMIT MESSAGE FIELDS:
109 Installed-by: username@hostname
110 Pre-commit-by: Real Name <username@mit.edu>
111 Upgraded-by: Real Name <username@mit.edu>
112 Migrated-by: Real Name <username@mit.edu>
113 Wizard-revision: abcdef1234567890
114 Wizard-args: /wizard/bin/wizard foo bar baz
118 Committer: Real Name <username@mit.edu>
119 Author: lockername locker <lockername@scripts.mit.edu>
123 - It is not expected or required for update scripts to exist for all
124 intervening versions that were present pre-migration; only for it
125 to work on the most recent migration.
127 - Currently all repositories are initialized with --shared, which
128 means they have basically ~no space footprint. However, it
129 also means that /mit/scripts/wizard/srv MUST NOT lose revs after
132 - Full fledged logging options. Namely:
133 x all loggers (delay implementing this until we actually have debug stmts)
135 - debug => loglevel = DEBUG
137 - default is WARNING (see below for exception)
138 - verbose => loglevel = INFO
139 x file logger (creates a dir and lots of little logfiles)
141 - log-file => loglevel = INFO
145 * Some parts of the infrastructure will not be touched, although I plan
146 on documenting them. Specifically, we will be keeping:
148 - parallel-find.pl, and the resulting
149 /mit/scripts/.htaccess/scripts/sec-tools/store/scriptslist
151 * The new procedure for generating an update is as follows:
152 (check out the mass-migration instructions for something in this spirit,
153 although uglier in some ways; A indicates the step /should/ be automated)
155 0. ssh into not-backward, temporarily give the daemon.scripts-security-upd
156 bits by blanching it on system:scripts-security-upd, and run parallel-find.pl
158 1. Have the Git repository and working copy for the project on hand.
160 2. Checkout the pristine branch
162 3. Run wizard `prepare-pristine APP-VERSION`
164 4. Checkout the master branch
166 5. [FOR EXISTING REPOSITORIES]
167 Merge the pristine branch in. Resolve any conflicts that our
168 patches have with new changes. Do NOT let Git auto-commit it
169 with --no-commit (otherwise, you want to git commit --amend
170 to keep our history clean
172 [FOR NEW REPOSITORIES]
173 Check if any patches are needed to make the application work
174 on Scripts (ideally, it shouldn't.) Run
175 `wizard prepare-new` to setup common filesets for our repositories.
177 6. Check if there are any special update procedures, and update
178 the wizard.app.APPNAME module accordingly (or create it, if
181 7. Run 'wizard prepare-config' on a scripts server while in a checkout
182 of this newest version. This will prepare a new version of the
183 configuration file based on the application's latest installer.
184 Manually merge back in any custom changes we may have made.
185 Check if any of the regular expressions need tweaking by inspecting
186 the configuration files for user-specific gunk, and modify
187 wizard.app.APPNAME accordingly.
189 8. Commit your changes, and tag as v1.2.3-scripts (or scripts2, if
190 you are amending an install without an upstream changes)
192 NOTE: These steps should be run on a scripts server
194 9. Test the new update procedure using our test scripts. See integration
195 tests for more information on how to do this.
197 http://scripts.mit.edu/wizard/testing.html#acceptance-tests
199 GET APPROVAL BEFORE PROCEEDING ANY FURTHER
201 NOTE: The following commands are to be run on not-backward.mit.edu.
202 You'll need to add daemon.scripts-security-upd to
203 scripts-security-upd to get bits to do this. Make sure you remove
204 these bits when you're done.
206 10. Run `wizard research appname`
207 which uses Git commands to check how many
208 working copies apply the change cleanly, and writes out a logfile
209 with the working copies that don't apply cleanly. It also tells
210 us about "corrupt" working copies, i.e. working copies that
211 have over a certain threshold of changes.
213 11. Run `wizard mass-upgrade appname`, which applies the update to all working
216 12. Run parallel-find.pl to update our inventory
218 * For mass importing into the repository, there are a few extra things:
220 * Many applications had patches associated with them. Be sure to
221 apply them, so later merges work better.
223 # the following operation might require -p1
224 patch -p0 < ../app-1.2.3/app-1.2.3.patch # [FIDDLY BIT]
226 * When running updates, if the patch has changed you will have to
227 do a special procedure for your merge:
229 git checkout pristine
230 # NOTE: Now, the tricky part (this is different from a real update)
231 git symbolic-ref HEAD refs/heads/master
232 # NOTE: Now, we think we're on the master branch, but we have
233 # pristine copy checked out
234 # NOTE: -p0 might need to be twiddled
235 patch -p0 < ../app-1.2.3/app-1.2.3.patch
237 # reconstitute .scripts directory
238 git checkout v1.2.2-scripts -- .scripts
240 # NOTE: Fake the merge
241 git rev-parse pristine > .git/MERGE_HEAD
243 You could also just try your luck with a manual merge using the patch
246 * The repository for a given application will contain the following files:
248 - The actual application's files, as from the official tarball
250 - A .scripts directory, with the intent of holding Scripts specific files
251 if they become necessary.
253 * .scripts/lock (generated) which locks an autoinstall during upgrade