This morning I saw the script had failed, and looked at the stdout dump to find the reason why. Lo and behold the reason became clear once I saw Vault's XML.
Code: Select all
<vault>
<!-- Too many errors were encountered downloading files. Some files may not have been updated -->
<result success="yes"/>'
</vault>
What sort of sick mind am I dealing with here????
No. If there were download errors then 'success' != 'yes' ... that's why my script failed, since the files weren't there to be built! I wonder how many times, thanks to this silent method of error reporting, that my build just happened to work OK but used stale files from previous GETs. It sure would explain a few things we've seen since moving to Vault.
Bah!