Getting the error “Can not parse generic service: CodeRunnerException: CodeRunner did not return object on requestActionInvocation”
Followed the advice in other posts, but it still keeps happening. Worked fine a few hours ago then stopped - at that point I had changed no code at all. I was redeploying because the service had started throwing odd 502 errors and I figured perhaps this would clean it up.
In any case I’ve rolled through all of the working versions from today and they don’t deploy. So either its a file I’ve messed up that is out of source control (not sure what, have NPM installed the whole thing several times) or something weird is happening
I attach the project code excluding the node_modules directory
Have you tried deploying it from your dev environment using the “npm run deploy” command? I believe all the dependant node modules must be in the scope of deployment.
Yes that’s how I’ve deployed it about 20 times today. Suddenly it doesn’t work anymore.
Kicking myself for deleting it, because it was working fine as far as I could tell apart from these 502 errors that appeared after it had been running fine for about 3 hours
Could you please zip up the entire directory (with all dependencies) and email to support@backendless.com ? If the file is too large, please upload to dropbox or google drive and email us the link (or post here - it is up to you).
We determined that the missing listing for the root directory is directly related to the presence of the .git folder. It looks like the request to show files in console ends up timing out before server prepares a response. The .git folder is rather large and it takes time to process it. We temporarily renamed .git to git-backup and you can see the files. So until a fix is available, there are the following options:
keep things as they are (this essentially disables git)
we can rename git-backup to .git, which will return git support, but will disable console listing.
Please let me know.
Ah ok, yeah a bunch of images - it’s got itself hunting them and their revisions for too long - understood. Hmmm. It is the easiest way for me to deploy and I have somec content authors doing stuff too who are now used to it.
I guess we’d better go with no console listing. I’d make a new repo but it took 9 commits for all of those images anyway. Though saying that I could just ignore the images folder from now. Create a new repo etc.
Thanks. I tried to change a couple of lines to see if I could stop trying to cache things that are probably too big. But the whole thing won’t deploy again…