I installed CF2016U14, then updated the connector, then verified server.xml and workers.properites had the shared secret, then verified the updates was installed in CF Administrator, and finally verified the sites were up and running.
A few minutes later, some of the sites went down. Looking on the server, the CF 2016 service just stops running. Accessing the site gives:
The service is unavailable.
Service Temporarily Unavailable!
The server is temporarily unable to service your request due to maintenance downtime or capacity problems. Please try again later.
This is happening on severla of our 5 servers (at least 3 so far).
Has anyone encountered this?
Copy link to clipboard
Can you get to the cf admin? That does not involve the updated connector at all. If it's running, then cf is up.
I'm that case, you still have some problem with the connector config.
Thanks for the response. I'm not 100% sure if I was able to access CF Admin while the issue was happening, but all 5 servers have been restarted and everything is up and without issue so far. If it happens again I'll be checking CF Admin and whatever logs I can find.
I compared against my backups of the server.xml and worker.properties files, and the workers.properties file is identical (except for the new, shared secret line), and the server.xml has the same tuning parameters in the connecor line.
As far as a rollback plan if this continues to be an issue, would uninstalling the updates and restoring the backed-up workers.properites and server.xml files do the trick?
I'll watch it over Friday and the weekend and post back if I find anything else out. I'll also post back if everything is good (fingers crossed) through the end of Monday.
Good to hear.
But sadly, no, on the roll back--unless you also backed up the isapi_redirect.dll (on windows) or the mod_jk.so file (for apache).
I have complained to Adobe that the wsconfig upgrade makes no backup of these connector files.
Most folks will NOT. have thought to take manual backups.
Thanks for the additional info. We have nightly backups of all of the servers, so we'd be able to restore anything we need, as long as we're able to identify what we need to restore (or we could roll back the whole server as a more drastic approach).
Still no issues, so I think it was a fluke, or Windows being dumb and thinking the service had failed when the patch installer stopped it.