update std
add copyright
add midnightbsd link
remove social stuff
add www
add description and license fields.
formatting dates.
add new api calls
add runs api
add blessed
add blessed
add util to list all port depends
turn off link
add top blockers
add top blockers to magus www interface
catch run ids
fix definition
error handling - dont die, just give 404s like google expects
make fast cgi compatible
add text message
add 404 errors for the ports list page
return an error if we cant connect. fix the bool column for restricted checks
add a copy of ready ports
type is enabled
list depends by type
query types
depends of port relationship
add type column
add type
expose type information
add keyword
add a new script to delete runs.
change styling
fix path
add compare runs to template
we use postgresql now
fix data type
fix includes
more explanation of magus
change padding
dont do vertical spacing
drop resposive on the iamge for now
add a timestamp so we get new data
make the font readable
cleanup markup
typo
fixup media tag
spacing fixes
fixup css
add btn-default
jumbotron replaces hero unit
first pass at bootstrap 3
fix text
try to deal with nav issues
use svnweb rather than cvsweb now
first pass at using restricted as a flag
add restricted column
add restricted in front end
oops field name is really name
order the list by name to make it easier
fix select
write it as we go
use vars
fix a bug
first pass at pg rewrite
increase license column
oops we use bootstrap 2.3.2
color code
update pass/fail/warn styles
add readyports view
no reason to order by name
migrate schema to postgres
fixup some properties.
Attempt checkout of svn repo into specific directory.
Update magus to use subversion roots instead. In this case, we switched CvsDir and CvsRoot to VcsRoot and VcsDir to be more generic. While here, fix the licenseit script
revert 1.8. md5 is used for verify of installed files, but sha256 is used for package verification. really should use sha for both
magus seems to use md5 for file has (why?)
sort the results
add compare feature
fix alignment at low resolution
add runlist
update styling
error styled
improve machine styling
make fluid
Start using twitter bootstrap on Magus website
add files ctfiv put local on stargazer years ago
comment out header as it causes errors in safari
make valid xhtml
add MidnightBSD text.
protocol ambiguous
Create a second mysql database connection so we don't have problems with the first result still getting held.
add load_depends function. It creates dependencies list for each package we add to the index for consumption by magus clients. This allows clients to predict what dependencies it needs to fetch prior to fetching the main package. Otherwise, we get a scenario where we fetch a package only to find out we need 5 more and then 5 more after that and so on.
Teach magus to print the worker # in proc title. This lets us know which process is going berserk.
use xz for compression
add .mkshrc to the list
Add hash code to packages table so we can verify files are ok. Newer versions of mysql require pthread.
Add partial edit functionality. This was work ctriv@ had been doing.
Mport packages end in .mport
Get mirror list from mysql instead of hard coding it. Remove www as we don't store that in magus anyway.
This isn't quite what chris had in mind, but set the alias to the origin for now. This allows a user to type a relative path off of mports to get a specific port without knowing the "friendly" package name. Add indexes for some columns to speed up searches on the client.
This is the first version of the new bless utility for magus/packages. This will generate a sqlite3 database containing an index and mirror list for packages. Next steps: handle os version and arch populate mirror list for mysql? fill in aliases
Perl was excluded. We need it for future magus runs.
system perl now..
likewise make it usr/bin/perl since we've got system perl
I think this should be /usr/bin/perl We make a symlink the port case and now that it could be installed as part of the OS...
rewrote the algorithm to not be recursive. MUCH better now.
Make the slave-datadir if it doesn't already exist.
Add audit directory for current. (should this be conditional?)
the new approach was fundementally flawed. Go back to the old way.
fix stupid typo bug.
Update to make /etc and /usr/share from src. We also inject a resolv.conf into /etc, as the one from source isn't empty/non-existant. Right now the only way to change resolv.conf is to edit this script, a way to configure this at run time is still needed.
clean up some sorting stuff.
Enable bzip2 compression -y flag to tar Add a newline after the warning so it's easier to read.
After our patch to cvsweb, directory urls need a trailing /
Move chroot init to be with the rest of the init code. Avoid an edge case with sigchld and db drops this way.
Fix various bugs found in the last run.
clean up signal stuff. Deal with child worker exiting 6 (db drop).
have the node script write a pid file. It's handy.
use logger.
Move a bunch of stuff to the child module. Still need to clean up logging, but this gives it some badly needed orginzation.
Fix bug with deleting the old mports dir. Fixed bug with package uploading.
Magus 2.1: Many old bugs fixed and replaced with new exciting bugs.
Try some basic locking to keep two or more workers from updating the machine's run at the same time.
when the script reloads itself, reload the version from the magus mports tree
include the worker ID in verbose reports.
make the cvs checkout more configurable
-j flag: run multiple instances in parallel.
Here's a quick an dirty utility I'm using to make the ftp symlinks. It requires MySQL client library (5.x used). This is ugly and the user/password pair shouldn't be hard coded (i replaced them in the code) To use, cd to the packages directory for the release in question and execute. (All and Latest should be subdirectories)
Try to recover from a dropped DB connection. Don't use /usr/mports for the loopback, set up a mports tree in /usr/magus/slave-data. Don't flush locks on a machine level. This is in anticipation of running more than one magus process on a machine.
add nuke_internals.pl and top_blockers.pl
Use the anchor to make the status display work with forwards and backwards buttons.
set umask to something more usefull before checking out.
delete lock link.
add support for deleting locks. I'll give you enough rope....
add delete_lock cgi.
change how the stats look on run pages.
if name is given as a column for search explicitly, don't LIKE it. This helps with the port agg view.
add category stuff, tweak port page.
don't drop untested if the user is doing something tricky in search.
close up cross-scripting hole.
colorize the depends link based on their status.
new colors.
Send the status for depends to the port template.
start of improved searching, can do things like "curl status:fail" or "run:12 status:pass" can't do things involving joins yet, so no "curl arch:i386 status:fail"
missed can_reset in one spot
only show reset links if the port belongs to an active run
move result-list.tmpl to port-list.tmpl only show reset links if the port belongs to an active run
remove delete result script. results are dead man. dead.
reset port machine
new stuff for machine page
add new template files.
add stuff for machine page.
varius cleanups and additions.
add runs table.
let the user hide the port list.
fix typo so that run links work.
add start of run page.
bug fix
note events differently. this should give timestamps and machine links.
start of work for mk II database. Not complete, but usable.
fix logging bug.
download depends from the right location.
Yet another typo. sigh.
add pkgname column.
bug/typo fix on log creation.
chmod the tarballs 664 after moving them to the ftp dir.
Don't die if the run's package dir already exists, that's ok.
Move tarballs to the run tarballs dir after they are made.
various cleanups. Most importantly, we refresh the port object in the parent after the child exits. This insures that package is uploaded in the port passed.
Die if downloading the run tarball fails. We might need/want a retry scheme in the future, but this will be enough for now.
bug fixes.
milestone 2
clean up the cvs cmnd for life on stargazer.
create the subdir in packages for each run.
Set the tree id when the tarball is made, just to simplify things.
Updated for second milestone.
schema 2.0. Many many changes. Result and port tables have merged. There is now an entry for each port for each run. Runs are per osversion and arch. events have replaced subresults.
new schema.
Add a warning about /mnt
Preserve permissions on tar operation to simulate a real system. Remove /var/crash/ and /var/audit as they should not be needed for ports builds.
Remove the unused task stuff. Don't try to remove the mport tree if it isn't there. Bug fix in the cvs command.
make the user confirm result deletion.
Add delete button.
Lable the search button.
Sync with currently deplayed version: now with searching!
add the in-progress image.
sync with our current working version.
bug fix.
Various bug fixes.
Add cvs update support. Fix typo in cluster. Cluster stuff still not ready for prime-time.
Left align text in a th. With a little padding this looks much on the new wider tables.
Changed how result details work. We now load them in the background on demand inside a results list.
Add support for skip subresults.
add alternating rows.
Header fixes.
Changed the stats to a ul. fiddled margins.
*** empty log message ***
switch to horizontal stats.
send the stats information to each page.
try to make stats on all pages
darker...
hover
"the font is too small"
more definitions.
a few style tweaks.
just add thead and tbody tags...helps with some user agents.
Add magus webapp.
Add task support.
Add tasks table. Fix column on the categories join table.
add halt and resume cluster. update the update_cluster.pl script.
Add the start of update_cluster, the ng indexer. This is just a stub file currently.
Make the log data a longtext.
Make the log messages a bit more useful
A few typo fixes.
Add logging. Make scp errors more verbose. Add command line flags. Fix SIGINT bug.
Quick little script to delete results from the db
Refactored a good chunk of code. Added logging. Added docs. Cleaned up exception handling. Still a few known bugs. SIGINT isn't quite handled correctly yet.
Add logs table.
Add some new vars.
The subresults table had result_id, changed to result.
First beta version.
Add master and slave scripts. Add schema. Add sample configuration file.
Add notes on the design of the port testing system (magus).
Add script for building the chroot envirement that ports are tested in.