New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Static server issues #7879
Static server issues #7879
Conversation
Still stuff not working. Looks like setting expert GMT_DATA_SERVER=static works differently (better?) than having gmtest.in do it.
Much harder to get |
Seems when tests are run in parallel there are hiccups where $GMT_DATA_SERVER is not detected and it defaults to oceania and it writes some files into .gmt/server and .gmt/cache. If I run the tests serially, one at the time then every remote file is saved under the static directory. I suspect something with sub-shells and some parts of scripts being run without $GMT_DATA_SERVER set. I dont know what might happen but clearly many files are being written to .gmt/server even though we have
in gmtest[.in]. Example of error message in the log:
So not sure how it selects .gmt/server here since it clearly knows we are using static (the message). There has to be either a remaining bug in the code or something with the exports. If I look at the log and see that ex27 failed for the same reason, I can do chest -R ex27 and it passes... |
Running Cmake prints
but when I run all the tests it only creates .gmt/server and .gmt/cache. No static dir is created. I added some print statements to gmtremote_init_paths and nothing is written to the log. How is this possible? Building again and this time I get a log with lots of
Yet things are saved in server and cache. Not sure what to think. |
Oh well, found my bug. Now I get consistently files in static.{server, cache} when either setting static in the environment or setting in in ConfigUserAdvanced.make. Please see if the latest update works for you, @seisman. |
I now get 30 failures only - all but two involve grid image differences. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Blindly approve because my Homebrew is broken so can't build GMT now.
Conclusions so far
It seems to be that this new line in gmtest.in
which becomes (when not set explicitly)
So that sets the default server probably because it is not set and thus defaults to oceania. Maybe the simplest is for the CI to simply have |
Following what you did, I also added Some important logs are pasted here:
As you can see, GMT now knows it should use the static server, but somehow it still tries to find files in So, it's likely there are still bugs, but I'm OK to approve this PR if you think the bugs are not worth your coding time. |
I follow your example to run just tut_11 and I get no errors and all is static and files downloaded to static. This is with no GMT_DATA_SERVER set in my environment. How can you and I get different results? I also broke the
So all seems good here? |
With If I comment out |
Strange. Here is my terminal (I am in the build directory). I comment out stuff in gmtest
and static is set in the terminal environment via export.
The test fails because the gm compare finds an rms exceeding the threshold, but it runs fine and grabs data from static. |
I feel it's just a minor issue and doesn't matter much. So maybe merge this PR and revisit the issue if it causes troubles. |
OK, merged. |
I got some errors with animation 2 because GMT searchs the remote data grids in the static sever. I think it is related to this PR. Later I can post more details. |
Don't know if related but the new planetary grids are still only in |
All the new stuff is still only on candidate. We release it to oceania and syncing hosts when 6.5 is released (same day/time). I did several of these bash aliases, e.g., alias set-static='export GMT_DATA_SERVER=static' and since I jump between servers for tests and try-outs I have to remember to do set-static before running tests and set-candidate before checking on Mars and mss etc. Works fine for me (?) |
But since the new datasets are already visible in https://www.generic-mapping-tools.org/remote-datasets/ why waiting to official 6.5? In my mix cmd-with-mingw those exports ( |
I usually use the candidate server. I suspect that GMT movie changes alone to the static server. |
Hm, yes those scripts started by system calls probably goes to the default server. So Oceania |
These changes simplifies the handling of USERDIR and CACHEDIR and now things seems to work if
Getting 32-33 failures if running one test at the time so that the downloads dont interfere.