Notifications
Clear all
Topic starter
When our JBOSS server restarts for updates or anything else, after the JBOSS service comes back up, if you try to login to CPS 12.3.0 using an LDAP account, a message in red writing says an Error Occurred, but you cannot see the full text of the message because it does not fit into that login window. You cannot login to the system.
If you use a non-LDAP account, defined within Centricity, you can login without any trouble, but that does not help most of our users.
Our VAR said this was caused by the QIE service (on another server) rapidly opening too many connections to the JBOSS service. I stopped the QIE service, restarted JBOSS, waited a few minutes, and started the QIE service back up again. LDAP sign-ons were working again, and if I do that manually it does seem to work.
I wrote a powershell script that does those steps, including the delays. The script is scheduled to run about 2 hours after the update window for the servers, but it does not work. It stops the services as expected, waits 10 minutes after JBOSS restarts before restarting QIE, but it does not fix the LDAP issue.
Has any one else had an issue with LDAP sign-ons not working if you restart your application server, and figured out a fix or workaround for it?
Posted : March 3, 2019 3:01 am
We just started seeing this since upgrading to 12.3.2. I was not aware of the QIE connection. I always shut down/ restart our dts/QIE server when I restart JBOSS. I suppose that is what is resolving our issue.
Not really help for you but certainly helps understand the issue.
Thanks,
Jack
Posted : March 3, 2019 3:46 am
I was told by support this was a known issue and that I have to just restart jboss multiple times until it works. For years I had weekly automated reboots of Jboss and SQL, just because it helped the system perform better but I stopped doing that after getting burned and thinking something major was wrong and calling clinic down on a Monday morning. They did not give a time table for a fix but we just learned to live with it. It seems like we have been getting it for a while, even before our 12.3.1 upgrade last year. I think we saw it in 12.2.x also.
What really stinks about it is you don't know if it is working until Jboss loads completely. I have had it not work for four iterations once but most of the time it is once or twice.
Mike Zavolas
Tallahassee Neurological Clinic
Posted : March 4, 2019 4:50 am
Could it be related to this issue which is in scope for SP3... We have a separate interop server and are experiencing this issue and have to bounce it again on occasion.
| CISDB00071035 | JBoss Interop server does not start cleanly in some cases | 43395 | Both | Yes |
Posted : March 7, 2019 5:05 am
We ran into this problem as soon as we upgraded to 12.2. We're now on 12.3 and I believe it still happens. We would run into it when we did our server reboots. What seems to work for us is to reboot the database server - then wait 15 minutes - then reboot the jboss server(s).
Steve
Posted : March 14, 2019 11:21 pm
Topic starter
As mentioned by Mike above, the problem with this glitch is that you cannot tell without manually logging in if LDAP is actually working or not.
I scripted this so that 15 minutes after our app server is restarted, it stops JBOSS, waits a minute, then shuts down QIE and DTS on the other server. Waits 2-3 minutes for that to finish, then starts JBOSS again. Waits 15 minutes, and then starts QIE and DTS.
This script is running, and I can see the log entries as the services shut down as expected on both servers, and start back up. I log return codes in the script, no issues. But it just doesn't work.
This started for me after we upgraded to 12.3.0. I had workarounds in place with 12.2, similar things, and never had an issue.
So updating/restarting the servers means I have to manually check LDAP logins (CPS logins work fine) and re-run my script until it comes up right. Truth is it's probably not a QIE problem. Just a random LDAP startup problem caused by some bad code, and simply restarting JBOSS repeatedly would probably fix it, too. In fact, it has in the past. It does seem to come up a little more reliably with QIE off, but it's pretty hard to know for sure. I might just be getting lucky on that particular JBOSS restart. Small sample size.
Posted : March 15, 2019 3:06 am
Excellent post. I got valuable information from here. Thank you!
Java Online Course
Posted : April 19, 2019 7:26 pm
We also have experienced this issue with EMR 9.12.2 and were rebooting JBoss 4-5 times with each regular maintenance cycle adding 30-40 minutes onto our downtime window. We were told by support to stop QIE before rebooting JBoss, wait for JBoss to come back up fully on our interops JBoss, then restart QIE. That has helped a lot. We were also seeing the .war deployments time out when JBoss was restarted so we were able with the help of support to extend the time limit for that so they deploy successfully. All of this started post AzureAD/ API implementation.
Hope this helps!
Posted : November 18, 2019 5:03 am
