STR: 1. Find a push with several Android Talos jobs that failed for random infrastructure reasons (this isn't hard, that's virtually every single push). 2. Quickly retrigger them all (say, by clicking on the first, clicking the blue +, then alternating between N and click to run through a half dozen in a dozen seconds) 3. Wait to find out the results of your retriggering, then, when one of them failed to retrigger, figure out which one it was, by counting if you were quick enough to get it to fail before any of the previous ones were cleared, or by throwing up your hands and just retriggering them all again if you don't have all the retrigger jobs visible and in order Alternate STR: 1. Find yourself with a need to see the Talos Dromaeo job for every platform over the course of two or three pushes, say because you need to know whether it actually ran, or got coalesced. 2. Start by hovering over each "T", until you discover that every one has the tooltip "Talos Perfomance opt" rather than "Talos Dromaeo opt" 3. Settle into the annoyance of "look over at the top right, click a letter, look down at the bottom left, find the name, look over at the top right, click a letter, look down at the bottom left, find the name, look over at the top right" It'll be annoying to do, and annoying to maintain as releng churns the names and the contents around, but we need to parse out actual names for Talos runs, not just Taloshood (while retaining the generic Talos Performance parsing for the inevitable additions and changes that we won't know and nobody will tell us about beforehand).
Actually, the annoying part is http://hg.mozilla.org/users/mstange_themasta.com/tinderboxpushlog/file/8f19c6e6af28/js/MachineResult.js#l42 where if you give them separate types, you have to either repeat the whole bunch, or do something funky, to get the right scrape filter.
T(Tp4m, Tpan, Tp4, ...) - can we have it with grouping like mochitests?
(In reply to Armen Zambrano G. [:armenzg] - Release Engineer from comment #3) > T(Tp4m, Tpan, Tp4, ...) - can we have it with grouping like mochitests? Sure, that would be fine by me. Whatever is possible for tbpl folks to do.
7 years ago
Summary: Give individual Talos runs their own names → Give individual Talos jobs their own unique names on tbpl
Created attachment 565153 [details] [diff] [review] patch This is running on http://tbpl.swatinem.de/ philor: feedback on the naming would be nice. This is missing Windows xperf and the android jobs. What should I do about them? The patch also takes care of the ordering (alphabetic) automatically
Comment on attachment 565153 [details] [diff] [review] patch r+ on the code, philor can choose the abbreviations
Attachment #565153 - Flags: review?(mstange) → review+
What to do about xperf is an interesting question: from the log, it looks pretty broken, though it's green, and it doesn't seem to produce any output (to us, though bug 679865 makes it sound like it invisibly reports to something I've never heard of). I sort of wonder whether the answer is "hide it and don't worry about how it displays, since whatever it is, it isn't meant for per-push developer consumption."
Created attachment 565776 [details] [diff] [review] more patch How's this? I anchored the regexes, because I didn't like the odds of something happening again like the way we were calling tpan just another tp. I'd rather see ts2 as a T until we adjust the regex than see tsomethingelse just disappear as a second ts. I threw in xperf as an x, figuring that it wouldn't be too hard to tell that it wasn't xpcshell and wasn't txpcshell, and that way we won't have to keep looking at it to be sure it isn't some newly added suite. This gets us everything except scroll and cold, which we only run on Mac on 3.6, which is well beyond the level of my disinterest. I only did tp4 because we still run it on try, as well as 3.6.
Attachment #565776 - Flags: review?(arpad.borsos)
And based on ten minutes of using it, I now desperately want this, and want badly to land it before anyone gets around to doing my deploy-to-prod bug so I can have it this week, but I desperately want it for within-a-push reasons. Anyone who thinks it will magically take away the need to fix bug 689625... good luck finding a way to use it for that.
Comment on attachment 565776 [details] [diff] [review] more patch Review of attachment 565776 [details] [diff] [review]: ----------------------------------------------------------------- Yes, the new regexes should be safer here.
Attachment #565776 - Flags: review?(arpad.borsos) → review+
Status: ASSIGNED → RESOLVED
Last Resolved: 7 years ago
Depends on: 693121
Resolution: --- → FIXED
Comment on attachment 565153 [details] [diff] [review] patch Shows how often I look at my review queue these days, doesn't it?
Product: Tree Management → Tree Management Graveyard
You need to log in before you can comment on or make changes to this bug.