Deployment broken?

Scriptcase 8 Bugs
#1

Hi

Been away for quite a while folks - but now I’m back :slight_smile: (Had to focus on Wordpress builds for a client almost exclusively for last 6 months)

Anyway - now I’m back - I’ve had to do a complete rebuild of my SC set up. SC was running on Ubuntu, but is now installed on Windows 10. Also using the PHP 5.6 version (8.1.041). So far so good. Sorted out all the path differences between linux and Windows (I think) and all apps generate without error. Likewise, the dev site seems to work as it should.

So, I went to deploy it all on Prod - oh dear.

I followed the typical deployment process keeping all the defaults, but pointing to my starting app. I chose to make a Zip. Clicked Next, and… Nothing. After a minute or so the page just dies. Tried it on different browsers in case that was the issue - same result.

I cleared down /tmp - same result.

I do see (when I tried again) that it gets to the point of producing an sc_pub_xxxxxxxxxxxxx folder - this looks like it contains all the apps, and maybe it’s the zipping of that folder that’s dying?

Any ideas people? Can I manually zip up that folder and carry on as a workaround?

In advance - thanks :slight_smile:

#2

How big is the project? If you deploy “selected apps” and choose a few, fails too?

#3

Hi Giu

It’s 40 apps and I have progressed a little since my post. It is definitely size related.

I tried these: [LIST=1]

  • 1 app with no common libs or CSS - zipped ok
  • 1 app with css only - zipped ok
  • 1 app with common lib only - zipped ok (but took a while)
  • 1 app with both common libs and CSS - failed! [/LIST] What I have done - but have yet to confirm works is - is to create my own zip.

    I repeated 2 and 3 above, and then did all apps with no css or common lib. From the /tmp library I took the 3 “sc_pub_xxxxxxx” folders and copied the contents of all 3 into a new fresh folder (replacing duplicates names as necessary). I then zipped that up. I will try later today to see if that deploys as it looks like the right format.

    So - it does seem size related when I try it as it should be - i.e. all apps and all the libs (at least for the first time). However, when I do that I get no message except sometimes it seems to time out with an “internal server error”. I did set up SC manually so I wonder if I have an apache setting that needs to be bigger? I’m no expert so I’m not sure?

    • #4

    Did you check these settings? http://php.net/manual/en/function.set-time-limit.php

    You may need to increase the timeout on your server. Because you said (but took a while) it made me think that it took longer then the default timeout setting.

    #5

    Thanks rr - I’ll have a look. My default_socket_timeout was upped to 300 from 60 (in php.ini), and my Timeout in Apache is 300. Same result :frowning:

    #6

    I assume you also did restart your server, otherwise setting it doesnt really do something…
    So can I assume that you deployed on a windows server? If so then this runs on your windows 10? I wonder which web server you are using then (uniform server (uniserverz) or xampp seem to work well).

    #7

    Hi rr - yes, I restart Apache whenever I change Apache or PHP config. Yes it’s Apache running on Windows 10. WAMP and XAMPP both use thread-safe versions of PHP which conflicts with Zend Guard Loader - so I did the installs of the WAMP stack manually and individually, using a non-thread-safe version of PHP. I did start off using SC’s own “full” install, but reverted to trashing it and starting over (I can’t recall why now - but that set up meant I couldn’t do something I needed).

    I can confirm that my workaround works - insofar I can deploy bits separately (as described in my earlier post) and then merged and zipped the results. So, whilst annoying, not a showstopper right now.

    #8

    Maybe this will help then… https://support.zend.com/hc/en-us/articles/206790567-Howto-Setting-Up-Apache-PHP-NTS-and-Guard-7-Loader-on-Windows

    All I did on Window Server 2008 was install uniserver set up oracle client and the path to reach the oracle client and a reboot and basically that was it.

    #9

    rr

    Thx - had pretty much done all that - I had even spun up a VM of the backup of my old Ubuntu and compared the apache and php configs of that to my current setup (nothing obvious there) - so far my setup works completely fine - it’s just the deployment when including the common libs that seems to cause it to barf. Not much in the logs either to indicate what’s upset it. I’ll keep trudging on…

    #10

    This is a tough one… I cant help you much there… You may get more info using the sysinternals suite from microsoft running (since you are on windows 10) whilst you are doing the generating. In this case procmon. But be aware that using these tools is pretty advanced. You can set up a filter to only monitor what php is doing but you have to count on several tens of thousands of lines.
    But you can detect errors if you know what to read for…
    This is the tool I use to find weird unexplainable errors tho…

    With processexplorer you can see if the php dll gets a timeout and dies on you (assuming that it does).
    This is more of a last resort way of trying to fix it tho… You may likely find a cause but it will take time to find it…

    You need to run the tools as Administrator!

    #11

    Thanks rr! That’s new stuff to little old me - but I will definitely look into it. Should I get any answers / resolution I will update in here.