6 - Certain classes of error will continually fail, so they should
7 put in a different "seen" file which also skips them, unless
8 we have some sort of gentle force
10 - Keep my sanity when upgrading 1000 installs
11 - Distinguish between errors(?)
12 - Custom merge algo: absolute php.ini symlinks to relative symlinks (this
13 does not seem to have been a problem in practice)
14 - Custom merge algo: check if it's got extra \r's in the file,
15 and dos2unix it if it does, before performing the merge
16 - `vos exa` in order to check what a person's quota is. We can
17 figure out roughly how big the upgrade is going to be by
18 doing a size comparison of the tars: `git pull` MUST NOT
19 fail, otherwise things are left conflicted, and not easy to fix.
20 - Prune -7 call errors and automatically reprocess them (with a
21 strike out counter of 3)--this requires better error parsing
22 - Snap-in conflict resolution teaching:
23 1. View the merge conflicts after doing a short run
24 2. Identify common merge conflicts
25 3. Copypaste the conflict markers to the application. Scrub
26 user-specific data; this may mean removing the entire
27 upper bit which is the user-version.
28 4. Specify which section to keep. /Usually/ this means
29 punting the new change, but if the top was specified
30 it means we get a little more flexibility. Try to
31 minimize wildcarding: those things need to be put into
32 subpatterns and then reconstituted into the output.
34 - Distinguish from logging and reporting (so we can easily send mail
36 - Logs aren't actually useful, /because/ most operations are idempotent.
37 Thus, scratch logfile and make our report files more useful: error.log
38 needs error information; we don't care too much about machinability.
39 All report files should be overwritten on the next run, since we like
40 using --limit to incrementally increase the number of things we run. Note
41 that if we add soft ignores, you /do/ lose information, so there needs
42 to be some way to also have the soft ignore report a "cached error"
43 - Report the identifier number at the beginning of all of the stdout logs
44 - Don't really care about having the name in the logfile name, but
45 have a lookup txt file
46 - Figure out a way of collecting blacklist data from .scripts/blacklisted
47 and aggregate it together
48 - Failed migrations should be wired to have wizard commands in them
49 automatically log to the relevant file. In addition, the seen file
50 should get updated when one of them gets fixed.
51 - Failed migration should report how many unmerged files there are
52 (so we can auto-punt if it's over a threshold)
54 - Let users use Wizard when ssh'ed into Scripts
55 - Make single user mass-migrate work when not logged in as root
57 - Make the rest of the world use Wizard
58 - Make parallel-find.pl use `sudo -u username git describe --tags`
59 to determine version. Make parallel-find.pl have this have greater
60 precedence. This also means, however, that we get
61 full mediawiki-1.2.3-2-abcdef names (Have patch, pending testing and commit)
62 - Make deployed installer use 'wizard install' /or/ do a migration
63 after doing a normal install (the latter makes it easier
66 - Pre-emptively check if daemon/scripts-security-upd
67 is not on scripts-security-upd list (/mit/moira/bin/blanche)
69 - Redo Wordpress conversion, with an eye for automating everything
70 possible (such as downloading the tarball and unpacking)
73 - Genericize callAsUser and drop_priviledges in shell
74 - Summary script should be more machine friendly, and should not
75 output summary charts when I increase specificity
76 - Summary script should do something intelligent when distinguishing
77 between old-style and new-style installs
80 - Don't use the scripts heuristics unless we're on scripts with the
81 AFS patch. Check with `fs sysname`
82 - Make 'wizard summary' generate nice pretty graphs of installs by date
83 (more histograms, will need to check actual .scripts-version files.)
84 - It should be able to handle installs like Django where there's a component
85 that gets installed in web_scripts and another directory that gets installed
87 - ACLs is a starting point for sending mail to users, but it has
88 several failure modes:
89 - Old maintainers who don't care who are still on the ACL
90 - Private AFS groups that aren't mailing lists and that we
92 A question is whether or not sending mail actually helps us:
93 many users will probably have to come back to us for help; many
94 other users won't care.
96 PULLING OUT CONFIGURATION FILES IN AN AUTOMATED MANNER
98 advancedpoll: Template file to fill out
99 django: Noodles of template files
100 gallery2: Multistage install process
101 joomla: Template file
102 mediawiki: One-step install process
103 phpbb: Multistage install process
104 phpical: Template file
107 wordpress: Multistage install process
109 PHILOSOPHY ABOUT LOGGING
111 Logging is most useful when performing a mass run. This
112 includes things such as mass-migration as well as when running
113 summary reports. An interesting property about mass-migration
114 or mass-upgrade, however, is that if they fail, they are
115 idempotent, so an individual case can be debugged simply running
116 the single-install equivalent with --debug on. (This, indeed,
117 may be easier to do than sifting through a logfile).
119 It is a different story when you are running a summary report:
120 you are primarily bound by your AFS cache and how quickly you can
121 iterate through all of the autoinstalls. Checking if a file
122 exists on a cold AFS cache may
123 take several minutes to perform; on a hot cache the same report
124 may take a mere 3 seconds. When you get to more computationally
125 expensive calculations, however, even having a hot AFS cache
126 is not enough to cut down your runtime.
128 There are certain calculations that someone may want to be
129 able to perform on manipulated data. As such, this data should
130 be cached on disk, if the process for extracting this data takes
131 a long time. Also, for usability sake, Wizard should generate
132 the common case reports.
134 Ensuring that machine parseable reports are made, and then making
135 the machinery to reframe this data, increases complexity. Therefore,
136 the recommendation is to assume that if you need to run iteratively,
137 you'll have a hot AFS cache at your fingerprints, and if that's not
138 fast enough, then cache the data.
140 COMMIT MESSAGE FIELDS:
142 Installed-by: username@hostname
143 Pre-commit-by: Real Name <username@mit.edu>
144 Upgraded-by: Real Name <username@mit.edu>
145 Migrated-by: Real Name <username@mit.edu>
146 Wizard-revision: abcdef1234567890
147 Wizard-args: /wizard/bin/wizard foo bar baz
151 Committer: Real Name <username@mit.edu>
152 Author: lockername locker <lockername@scripts.mit.edu>
156 - It is not expected or required for update scripts to exist for all
157 intervening versions that were present pre-migration; only for it
158 to work on the most recent migration.
160 - Currently all repositories are initialized with --shared, which
161 means they have basically ~no space footprint. However, it
162 also means that /mit/scripts/wizard/srv MUST NOT lose revs after
165 - Full fledged logging options. Namely:
166 x all loggers (delay implementing this until we actually have debug stmts)
168 - debug => loglevel = DEBUG
170 - default is WARNING (see below for exception)
171 - verbose => loglevel = INFO
172 x file logger (creates a dir and lots of little logfiles)
174 - log-file => loglevel = INFO
178 * Some parts of the infrastructure will not be touched, although I plan
179 on documenting them. Specifically, we will be keeping:
181 - parallel-find.pl, and the resulting
182 /mit/scripts/.htaccess/scripts/sec-tools/store/scriptslist
184 * The new procedure for generating an update is as follows:
185 (check out the mass-migration instructions for something in this spirit,
186 although uglier in some ways; A indicates the step /should/ be automated)
188 0. ssh into not-backward, temporarily give the daemon.scripts-security-upd
189 bits by blanching it on system:scripts-security-upd, and run parallel-find.pl
191 1. Have the Git repository and working copy for the project on hand.
193 /- wizard prepare-pristine --
195 A 2. Checkout the pristine branch
197 A 3. Remove all files from the working copy. Use `wipe-working-dir`
199 A 4. Download the new tarball
201 A 5. Extract the tarball over the working copy (`cp -R a/. b` works well,
202 remember that the working copy is empty; this needs some intelligent
205 A 6. Check for empty directories and add stub files as necessary.
206 Use `preserve-empty-dir`
210 7. Git add it all, and then commit as a new pristine version (v1.2.3)
212 8. Checkout the master branch
214 9. [FOR EXISTING REPOSITORIES]
215 Merge the pristine branch in. Resolve any conflicts that our
216 patches have with new changes. Do NOT let Git auto-commit it
217 with --no-commit (otherwise, you want to git commit --amend
218 to keep our history clean
220 [FOR NEW REPOSITORIES]
221 Check if any patches are needed to make the application work
222 on Scripts (ideally, it shouldn't.
224 /- wizard prepare-new --
226 Currently not used for anything besides parallel-find.pl, but
227 we reserve the right to place files in here in the future.
230 A echo "Deny from all" > .scripts/.htaccess
234 10. Check if there are any special update procedures, and update
235 the wizard.app.APPNAME module accordingly (or create it, if
238 11. Run 'wizard prepare-config' on a scripts server while in a checkout
239 of this newest version. This will prepare a new version of the
240 configuration file based on the application's latest installer.
241 Manually merge back in any custom changes we may have made.
242 Check if any of the regular expressions need tweaking by inspecting
243 the configuration files for user-specific gunk, and modify
244 wizard.app.APPNAME accordingly.
246 12. Commit your changes, and tag as v1.2.3-scripts (or scripts2, if
247 you are amending an install without an upstream changes)
249 NOTE: These steps should be run on a scripts server
251 13. Test the new update procedure using our test scripts. See integration
252 tests for more information on how to do this.
254 http://scripts.mit.edu/wizard/testing.html#acceptance-tests
256 GET APPROVAL BEFORE PROCEEDING ANY FURTHER
258 NOTE: The following commands are to be run on not-backward.mit.edu.
259 You'll need to add daemon.scripts-security-upd to
260 scripts-security-upd to get bits to do this. Make sure you remove
261 these bits when you're done.
263 A 14. Run `wizard research appname`
264 which uses Git commands to check how many
265 working copies apply the change cleanly, and writes out a logfile
266 with the working copies that don't apply cleanly. It also tells
267 us about "corrupt" working copies, i.e. working copies that
268 have over a certain threshold of changes.
270 A 15. Run `wizard mass-upgrade appname`, which applies the update to all working
271 copies possible, and sends mail to users to whom the working copy
272 did not apply cleanly.
274 16. Run parallel-find.pl to update our inventory
276 * For mass importing into the repository, there are a few extra things:
278 * Many applications had patches associated with them. Be sure to
279 apply them, so later merges work better.
281 # the following operation might require -p1
282 patch -p0 < ../app-1.2.3/app-1.2.3.patch # [FIDDLY BIT]
284 * When running updates, if the patch has changed you will have to
285 do a special procedure for your merge:
287 git checkout pristine
288 # NOTE: Now, the tricky part (this is different from a real update)
289 git symbolic-ref HEAD refs/heads/master
290 # NOTE: Now, we think we're on the master branch, but we have
291 # pristine copy checked out
292 # NOTE: -p0 might need to be twiddled
293 patch -p0 < ../app-1.2.3/app-1.2.3.patch
295 # reconstitute .scripts directory
296 git checkout v1.2.2-scripts -- .scripts
298 # NOTE: Fake the merge
299 git rev-parse pristine > .git/MERGE_HEAD
301 You could also just try your luck with a manual merge using the patch
304 * The repository for a given application will contain the following files:
306 - The actual application's files, as from the official tarball
308 - A .scripts directory, with the intent of holding Scripts specific files
309 if they become necessary.
311 * .scripts/lock (generated) which locks an autoinstall during upgrade