Perfomatic:Rearchitecture: Difference between revisions
Jump to navigation
Jump to search
| Line 16: | Line 16: | ||
=== Random Ideas === | === Random Ideas === | ||
* bulk and collect.cgi are the same except bulk can handle multiple data points. They should be combined. | * bulk and collect.cgi are the same except bulk can handle multiple data points. They should be combined. - [https://bugzilla.mozilla.org/show_bug.cgi?id=456529 bug 456529] | ||
* Need to connect data points on continuous graph to discrete - bug 438745 | * Need to connect data points on continuous graph to discrete - [https://bugzilla.mozilla.org/show_bug.cgi?id=438745 bug 438745] | ||
* Need overlay of points on continuous graph that shows buildid and changeset id - [https://bugzilla.mozilla.org/show_bug.cgi?id=438745 bug 431372] | |||
* Need overlay of points on continuous graph that shows buildid and changeset id - bug 431372 | |||
* Too much data to query averages and look for regression | * Too much data to query averages and look for regression | ||
* Don't want developers to know machine names | * Don't want developers to know machine names | ||
* Dashboard that shows 3 os's and color for status | * Dashboard that shows 3 os's and color for status - [https://bugzilla.mozilla.org/show_bug.cgi?id=456567 bug 456567] | ||
* Need a way to show/test health of build and health of tests | * Need a way to show/test health of build and health of tests | ||
* Need to test for machine agreement (3 machines run same test, if 1 is way off, that machine is mostly likely broken). | * Need to test for machine agreement (3 machines run same test, if 1 is way off, that machine is mostly likely broken). - [https://bugzilla.mozilla.org/show_bug.cgi?id=457780 bug 457780] | ||
* Need to remove tests for unmonitored branches/builds. | * Need to remove tests for unmonitored branches/builds. - [https://bugzilla.mozilla.org/show_bug.cgi?id=457782 bug 457782] | ||
* Waterfall doesn't show test results, just that tests ran, which is not a good indicator of build health. | * Waterfall doesn't show test results, just that tests ran, which is not a good indicator of build health. | ||
* Need to stop using machine names for test/build/os | * Need to stop using machine names for test/build/os | ||
Latest revision as of 00:13, 30 September 2008
Backend
- New DB schema
- InnoDB
- Add triggers to recalculate averages - bug 456530
- Add stored procedure to remove old data - bug 456532
- Add memcached or sqlite for frequently accessed data
- Depends on how slow the new db is
- Send less data for continuous graphs - bug 456534
Frontend
- Add more search criteria - bug 456536
- Show machine info per test
- Don't show tests for discrete data
- Don't iterate over datasets
- Don't parse test info for test information
Random Ideas
- bulk and collect.cgi are the same except bulk can handle multiple data points. They should be combined. - bug 456529
- Need to connect data points on continuous graph to discrete - bug 438745
- Need overlay of points on continuous graph that shows buildid and changeset id - bug 431372
- Too much data to query averages and look for regression
- Don't want developers to know machine names
- Dashboard that shows 3 os's and color for status - bug 456567
- Need a way to show/test health of build and health of tests
- Need to test for machine agreement (3 machines run same test, if 1 is way off, that machine is mostly likely broken). - bug 457780
- Need to remove tests for unmonitored branches/builds. - bug 457782
- Waterfall doesn't show test results, just that tests ran, which is not a good indicator of build health.
- Need to stop using machine names for test/build/os
- Win & linux throttled, mac not. This makes comparing Win & Linux to Mac useless.
- Need to figure out what tests we really need
- Terminology is confusing
- Date and time have differences depending on type of graph
- Running Talos on a dev machine, you can't graph anything or get useable numbers