I suspect that I may have some corruption in my SGD object database. Last weekend I upgraded our two-server array from version 4.62 to 5.0. The procedures for upgrading an array were followed properly. After the upgrades, we are experiencing strange things.
We also upgraded two other standalone servers (no array) and they do not have any issues.
This a a Solaris array.
Here are a few examples:
1. Application launches are timing out trying to connect from the SGD servers to the application servers.
2. Unexplained increased network load.
3. Problems with the password caches:
#tarantella passcache list
The passcache list command failed for an unknown reason.
4. Sometimes an application launch will display a log in prompt from the application server, with garbage in the username field.
Is there any way to verify/troubleshoot/repair the integrity of the SGD object database?
Message was edited by: Jeffro
Oracle support acknowledged corruption with the password cache. This was resolved with the following process:
Stop SGD on both SGD servers:
# tarantella stop
Backup current passcache:
# cd /opt/tarantella/var/info/
# cp -pr tier3cache tier3cache.bak
Remove all passcache entries
# cd tier3cache
# rm *
# tarantella start