To unsubscribe from this group and stop receiving emails from it, send an email to vcap-dev+u...@cloudfoundry.org.
To unsubscribe from this group and stop receiving emails from it, send an email to vcap-dev+u...@cloudfoundry.org.
since we have a “rename” command in the CLI and can do so without changing the GUID of the app, one of the things i had talked to scott about was instead of using the concept of blue/green, we could use an app suffix name convention. so when pushing a new app to the env in a zero-downtime way, you would have a concept of a production route, say, myapp.example.com. let’s take a look at the suffixes and examples in this scenario:
APPNAME-current - the live in production version mapped to the production route
APPNAME-new - the new version of the app that is intended to take the place of the production app
APPNAME-old - the previous production version of the app
i’m not attached to the suffix names in these examples, it could be next/previous instead of old/new or something else.
time0, the apps in a space look like this:
APPNAME-current - running and mapped to myapp.example.com
APPNAME-old - not running, still mapped to myapp.example.com
time1, the apps in a space look like this:
APPNAME-current - running and mapped to myapp.example.com
APPNAME-old - the previous production version of the app, not running, still mapped to myapp.example.com
APPNAME-new - pushed to CF with a route like myapp-next.example.com and is running at the end of this step to continue
time2, the apps in a space look like this:
APPNAME-current - running and mapped to myapp.example.com
APPNAME-old - the previous production version of the app, not running, still mapped to myapp.example.com
APPNAME-new - once running, this version now gets mapped to myapp.example.com
time3, the apps in a space look like this:
APPNAME-current - running and mapped to myapp.example.com
APPNAME-old - this version is deleted or renamed to something else
APPNAME-new - running and mapped to myapp.example.com
time4, the apps in a space look like this:
APPNAME-old - app is renamed, still running and mapped to myapp.example.com
APPNAME-new - running and mapped to myapp.example.com
time5, the apps in a space look like this:
APPNAME-old - app is stopped, and still mapped to myapp.example.com
APPNAME-current - app is renamed, running and mapped to myapp.example.com
you could even keep several copies of the older apps like:
APPNAME-old-1
APPNAME-old-2
APPNAME-old-3
what do you think of this approach therefore each deploy takes the same flow instead of swapping blue/green and having to remember which is which.
great job on the version 2 of this buildpack.
I wanted to test Spring Insight. Is there any document that describes how we can get Spring Insights working within PCF?
Hey,
great job on the version 2 of this buildpack.
I wanted to test Spring Insight. Is there any document that describes how we can get Spring Insights working within PCF?
-- Matthias
To unsubscribe from this group and stop receiving emails from it, send an email to vcap-dev+u...@cloudfoundry.org.
* I know you said that the v2 breakage was not taken lightly. I just want to emphasize that it was a pain for us and we hope we won't have to do it again for a long while. :)
* For our unit tests we use the same test helpers you use in your tests. One of the difficult parts of the upgrade is we had to re-write the tests for our extensions in addition to rewriting the actual logic. Do you consider your test helpers a public api as well? It would be nice if you did.
Our DEAs don't have access to the internet. Instead of relying on the buildpack-cache for application binaries we are now syncing an internal repository with s3://download.pivotal.io. We then maintain our own index.yml files to make visible versions that we wish to expose to our customers when we are ready to expose them.
Anyway, just thought I'd make you aware in case there is an issue with us doing that. :)
In master, we've broken up the Tomcat component into something more modular and maintainable and that is likely to hurt you guys a bit (which sucks considering how much work you've done on the buildpack recently). If you need some help updating your buildpack to cope with this, send me an email off-list.
To unsubscribe from this group and stop receiving emails from it, send an email to vcap-dev+u...@cloudfoundry.org.
2. App dynamics sends up all Environment variables. This is a security risk for us sending VCAP_SERVICES to app dynamics. So we have some custom hacks to not send those up. Once app dynamics provides official support for filtering what environment variables get sent to app dynamics there is probably some potential for enhancement here.
3. We support a custom app-agent-config.xml file in "resources". This is something that would probably be generally useful but it was such a small amount of work I haven't gotten around to it.
Ah, interesting. Do you know if they're actively working on this filtering? If not, it's probably something that we can push on a bit. Seems to be a pretty big security risk in a PaaS environment.
We have requested it a month or 2 ago. It sounded like others had requested it and that they were planning to fix it. However, we haven't checked on the status lately.