we have our main branch Caplin-Trader-Main. Every two weeks we release. To release we create a branch of main called CaplinTrader-1.4.6 for example. The we setup a new pipeline in cruise for CaplinTrader-1.4.6 to build the branch of code. There is often a round of bug fixes or feature removals so people do check into the branch.

Every successful build we run a pipeline containing performance tests which can take their feed from either the branch or main, but for these tests to run it needs to know which one triggered it. We currently have to manually change the name of this var in the code every time we branch.

The value of the dependency label is very useful to us but the fact the name changes means we would have to keep modifying code to keep up with the name changes. which is annoying and unecessary.

does this make things any clearer?

This error has just reoccurred. I installed the Cruise Server Version: 1.2(6278-05138ca7feda) you sent me. Hopefully we have some logs that highlight the problem. What Qiao, what do you want me to send in?

thanks

toby

a bit hacky but managed to get changelist number from the console log on the server

url = ‘http://’config[‘cruiseserverurl’]’/cruise/files/’config[‘pipeline’]’/’config[‘pipelinecounter’]’/installer/LATEST/makeInstaller/cruise-output/console.log’
f = urllib.urlopen(url)
consoleLog = f.read()

thanks for confirming that.
As Tim (tlittle) points out the docs could do with an update. This was my first stumbling block and will be many other peoples too

Thanks for the reply Qiao

The python script i am using takes the artifact list in json served by cruise and then iterates over it downloading each file and mirroring the folder structure.

def getFiles(artifacts, filePaths): for artifact in artifacts: for path in filePaths: if artifact[‘url’].find(path) > 0: if artifact[‘type’] == ‘file’: print ‘Getting file ’, artifact[‘url’] # map the url to a file path based on common root path target = convertUrlToPath(artifact[‘url’]) # get the numbe rof letters in the file name and add one for the slash filenameLen = len(os.path.basename(target)) +1 # Now we know the length of the filename and slash trim this from path target = target[0:-filenameLen] #shouldn’t need to check if the dir exists as it should always hit the parent folder first int he json print ‘Change dir ’, target os.chdir(target) try: f = urllib.urlretrieve(artifact[‘url’]) dest = os.path.join( os.getcwd(), artifact[‘name’]) shutil.copy(f⁰, dest) print ‘SUCCESS copied to ’, dest except: print ‘Error: Could not download file: ’,artifact[‘url’]

elif artifact['type'] == 'folder':
    #need to turn the cruise url into a suitable file path
    target = convertUrlToPath(artifact['url'])

1. now we know what directory we need make sure it exists if not os.path.exists(target): mkdir(target)

os.chdir(target)
getFiles(artifact['files'], filePaths)

Like i said i don’t have any precise timings on the download speed of each file. I will add this in a bit.

My colleague has been looking at using fetchartifact and is experiencing similar issues. I have asked him to post up his findings at some point today.

thanks

toby

ah i see the docs have been updated.

Thanks Qiao

Here is my server info:
Cruise Server Version: 1.2(6244-435fc6c2823d)
JVM version: 11.0-b16
OS Information: Linux 2.6.9-67.0.15.plus.c4.VMware
Usable space in artifacts repository: 34287 Mb
Database schema version: 29
License expiry date: Fri Feb 27 00:00:00 GMT 2009
Number of agents Cruise is licensed for: 5

As you can see we are using 1.2 and as i mentioned we do use VM’s, but have not had file transfer issues when using file shares or other methods (ftp).

I need to do some more precise timings which i will do when i get my pipeline working.
We don’t have any anti virus on our Linux machines.