PDQ Pull Method for large deployment
We've been using the pull method with DFS w/ replication of installers to be able to use one pdq deploy and inventory setup across multiple buildings.
It has been working great except for "large" deployments where the installer file size (or all included files for the installer) are large (anywhere over a couple hundred MB's). Deployments such as this which work flawlessly with the push method seem to hang for pull method. They sit at "copying" forever until timing out (no % is ever displayed and the machines that are supposed to be pulling have no to little activity on them when it is sitting there). It seems that if the deployment file size is "large" that the clients or for some reason pdq deploy stops the pull copy and hangs.
Is there some limitation of file size hidden somewhere in the pull method?
Comments
Anyone else experiencing this? I can re-create it every single time with many different installers so it doesn't seem to be anything to do with one installer.
When the Pull copy method is used the files don't start copying right away. All of the files that are going to be pulled down are indexed in a file called files.ini in \\TargetComputer\ADMIN$\PDQDeployRunner\<num>
Depending on the number of files that are going to be pulled down this can take a while. For Office 2003, for example, there are over 4,200 files so just indexing the files takes around 8 minutes. This indexing is important because PDQ Deploy needs to let the Runner process know which files to pull.
During this step of indexing the files we will change the status from Copying to Indexing or Preparing or something more intuitive.
Here is a screenshot of files.ini as it is being written in \\Randy\ADMIN$\PDQDeployRunner\1
Here is a screenshot of the opened file. You can see the indexed files.
I forgot to mention that we will make this step much faster than it currently is.
Ah, that definitely makes sense. So when it's sitting there not copying for 15 minutes via pull method it technically is; it's just still on step 1 of indexing the files before copying. Typically I aborted after 10-15 minutes of them sitting there since they weren't copying or showing any signs of moving along. I think I'll try it again and watch the index file build on one to make sure it's going and see how it goes after that.
Any idea when the new indexing method will be introduced?
Thanks!
My advice in this case is to create a selfextracting archive, which you can use like an installer.
Advantages - you've got one smaller file and additional verification option. (file checksum)