Add a comment for disabling errexit, try to debug appveyor cache #41157
Conversation
aidanhs
changed the title
src/ci/init_repo.sh
Outdated
| @@ -41,6 +41,8 @@ if [ ! -f "$cache_valid_file" ]; then | |||
| rm -rf "$CACHE_DIR" | |||
| mkdir "$CACHE_DIR" | |||
| else | |||
| # Ignore errors while gathering information about the possible brokenness | |||
| # of the git repo since our gathered info will well us something is wrong | |||
appveyor.yml
Outdated
| - cat %CD%\sccache.log | ||
| - cat C:\Users\appveyor\AppData\Local\Temp\1\build-cache-logs\*.log | ||
| - cat %CD%\sccache.log || exit 0 | ||
| - cat C:\Users\appveyor\AppData\Local\Temp\1\build-cache-logs\*.log || exit 0 |
There was a problem hiding this comment.
Is this cat even useful here? Since the build itself may succeed, it's probably rarely execute.
There was a problem hiding this comment.
To be honest I'm relying on the ar.exe errors to let me read this log (if it rarely executes because the ar.exe errors vanish, I wouldn't see that as the end of the world!). I'll probably remove it again once it's all fixed.
There was a problem hiding this comment.
In fact, I'll move it elsewhere and then we can just try it rather than merging.
aidanhs
force-pushed
the
aphs-debug-appveyor-cache
branch
2 times, most recently
from
40c2bb2 to
66ccc6f
Compare
|
@TimNN ok, think this is ready for a try (not an r+!) and should be pretty quick to run. |
|
Unfortunately try doesn't work for rust-lang/rust :( Could you just add an exit 1 or something to the appveyor script? |
aidanhs
force-pushed
the
aphs-debug-appveyor-cache
branch
2 times, most recently
from
1dce93d to
c26c259
Compare
|
@TimNN pushed a deliberate error to both travis and appveyor so this doesn't consume too much CPU time. I assume it's possible to r+ even if the travis PR test has failed? |
|
@bors r+ Lets see how this goes. |
|
📌 Commit c26c259 has been approved by |
|
⌛ Testing commit c26c259 with merge 37aa503... |
|
💔 Test failed - status-appveyor |
|
@bors r- |
aidanhs
force-pushed
the
aphs-debug-appveyor-cache
branch
from
c26c259 to
2b0652c
Compare
|
Oh the joys of globs in cmd.exe. I've now wrapped it in bash instead. |
|
@bors r+ |
|
📌 Commit 2b0652c has been approved by |
|
⌛ Testing commit 2b0652c with merge 1e1071e... |
|
💔 Test failed - status-appveyor |
Well that's disturbing. Maybe I'd better ask the appveyor people about this. |
aidanhs
force-pushed
the
aphs-debug-appveyor-cache
branch
from
2b0652c to
bb84358
Compare
|
@TimNN I've done my debugging, so for now I've updated this PR with the comment and a removal of a useless cat (since the glob doesn't apply). The |
shepmaster
added
the
S-waiting-on-review
|
@bors r+ rollup |
|
📌 Commit bb84358 has been approved by |
TimNN
added
S-waiting-on-bors
…r=TimNN Add a comment for disabling errexit, try to debug appveyor cache Comments added as requested. Also, lets add some cache debugging to appveyor. I *think* this is how to ignore errors in cmd.exe (and I did try it on my own machine), but I'm not 100% sure how appveyor runs them. In the worst case it'll fail but I guess that isn't the end of the world since the build has already failed by this point. r? @TimNN
Add a comment for disabling errexit, try to debug appveyor cache Comments added as requested. Also, lets add some cache debugging to appveyor. I *think* this is how to ignore errors in cmd.exe (and I did try it on my own machine), but I'm not 100% sure how appveyor runs them. In the worst case it'll fail but I guess that isn't the end of the world since the build has already failed by this point. r? @TimNN
|
☀️ Test successful - status-appveyor, status-travis |
Comments added as requested.
Also, lets add some cache debugging to appveyor. I think this is how to ignore errors in cmd.exe (and I did try it on my own machine), but I'm not 100% sure how appveyor runs them. In the worst case it'll fail but I guess that isn't the end of the world since the build has already failed by this point.
r? @TimNN