In addition to my hosts file issue, when I tried starting my application server for the first time the other day, I ran into an error where everything started except for the JLS which is clearly kind of important. Everything looked fine and the configuration clearly worked on other servers. However, the problem turned out that when I set the application server Jolt settings I entered the machine name which was called 33psapp1. Since the server name starts with a number it gets all upset and thinks that I have entered an IP address when clearly I haven’t. So changing the name to the IP address solves the problem.

You can’t use a machine name that starts with a number in the psappsrv.cfg configuration file. Just simply enter the IP address.

The other day I was configuring a new production system for a client and I ran into something I have never seen before. I was trying to test the SSL certificates on the new servers and since the old servers are still running in production I can’t change the DNS entries so I simply fired up UltraEdit and modified my hosts file to have the new server IP point to the new server SSL host name. This always works for me, except this time. I checked the logs and I am seeing all sorts of odd SSL revoked errors in the logs and I am seeing the OLD IP address instead of the new IP address.

I am completely puzzled, all I can see is that I am not picking up the new IP address from the hosts file. After running this by one of the server administrators he tells me that he has seen the problem before, and that I need to disable the DNS Client Service.

DNS Client Service: The DNS Client service is the client component that resolves and caches Domain Name System (DNS) domain names. When the DNS Client service receives a request to resolve a DNS name that it does not contain in its cache, it queries an assigned DNS server for an IP address for the name. If the DNS Client service receives the requested address, it stores the name and address in its cache to resolve future requests without having to query the DNS server. All computers that use DNS to resolve domain names (including DNS servers and domain controllers) use the DNS Client service for this purpose.

Sure enough once disabled, the host file is read and we are able to test the SSL certificate without an issue. I have done some reading on this and I am not completely sure that disabling the service is necessary, however, it did work in this case. From what I read it maybe a problem with the registry or with security on the hosts file. Some people were able to resolve the issue by simply deleting the hosts file and creating a new one which changed the security permissions on the hosts file, others have reported that there is a policy item and the registry is point to an incorrect policy.

Irregardless, for my testing needs, stopping the service and updating the hosts file then validating that the SSL certificates were in place was sufficient.

I found an odd issue today when trying to start up an PeopleSoft Environment Hub Agent. The environment uses PeopleTools 8.53 in a decoupled configuration. When running the StartAgent.sh file it was reporting an error that it could not find the binary psae to run for the validation/crawl. It was saying that it could not be found in null/bin. This seems weird because everything is setup correctly as far as environment variables go.

With a little digging around it turned out that the peopletools.properties file in the PS_CFG_HOME had an incorrect installation path. Once I changed that the crawl started working without an issue. The peopletools.properties file in the PS_HOME was correct, so just make sure you check all the homes.

Well, I was just on Oracle Support and was downloading the new Finance Image 7 for 9.2 and it is now 9 separate files. This process is just going to grow more and more painful as time goes on, but it is what it is (for now).

So, I started to download each file and when I got to the 8 file, it will download no more. This is because the first download was complete and the other six was the maximum number of open connections allowed within the Firefox browser. However, a little searching revealed a quick fix to this maximum number of open connections per server:

Open up firefox – type about:config in the Firefox address bar
Look for: network.http.max-persistent-connections-per-server
I changed mine from 6 to 10, and was able to download all the files for the Image 7 at the same time.!

Oh the little things that make life easier.

I have to say I have been using the PUM now for several upgrades and patch sets and everytime I get a new image I end up beating my head against the wall trying to figure out little issues.

A common problem, I have seen is the initial load of the PUM is fine and if you only use it once you might never see any issues, however, on subsequent loads if you did not install the network connections configuration right, you might be in for a wild ride.

Today, I ran into an image that another person installed and when starting up the database and listener would not start, which means the web and app are going to be down as well. However, this one was even more special because I got some really odd errors including this one:

ORA-00000: normal, successful completion

Now maybe, I am a little dense but this is an error message, so why is it saying normal and successful completion and it is an error message. I found a few suggestions on my handy dandy Google search that suggested to look at the ORACLE_HOME, ORACLE_SID and ORACLE_BASE variables as well as to double check the hosts file. Well, before I start anything I usually have my ORACLE variables well in hand, so I checked the hosts file and found that all the references I had to localhost were not resolving because there was no entry in the hosts file. That seemed to resolve the situation so that the database would actually start, but I was still having major headaches because everything was pointing to localhost and localhost.localdomain and there was just a ton of resolution problems.

I ended up change the tnsnames.ora file which for an image is under the /etc/tnsnames.ora location the one under $ORACLE_HOME/network/admin has no effect. Once I changed the tnsnames.ora file and the listener.ora file to the actual ip address, I also added a local_listener variable to the init{SID}.ora file for the environment. This solved the database issues, however, I had to also hard code the IP address for the JSL (Jolt) for the application server in order to get it to start, and for some final insult to injury, I ended up re-installing the PIA in order to fix all the invalid references.

For working with the PUM, all database functions are under the user: oracle, all PeopleSoft functions are under the user: psadm2.
Jolt Port is: 9000, webserver http is: 8000, default pia site is: ps, and if you need the IP address assigned to the Virtual box image, when you log in as root, type: ifconfig

Happy upgrading and patching.