Hosted Servers Best Practices

Currently my employer's Infor Lawson 10x system is hosted in Amazon's AWS. We do our own administration. We use SQL Server. I cannot answer your SQL questions, sorry, as we have done nothing special. Our data is relatively light so we always have good performance.

I can answer your other questions. Rebooting in the AWS environment should technically not be necessary, but we found as we ran in test mode during this past winter that after about 10 days of continual up-time, funky things began to happen. There's no need to reboot, however we now have a scripted outage period from 2:00 AM to 2:15 AM on Saturday mornings, and since then we have had no issues. The outage is just two scripts, a Stop script and a Start script, which take all Lawson services down and then back up, including SQL Server, WebSphere, etc. Everything. But the AWS servers and the OS are left alone.

As far as CTP patching goes, I have found the scripts to take Lawson down and back up again most helpful. Before I start a CTP, I run the Stop script. This kills all background processes that seem to lurk in ways that screw up dbreorg. Then I bring the entire system back up with the Start script. Then I kill just the last top-most Websphere process. No one can get to Lawson after that, except through LID, which is now forbidden for the regular users. Sometimes I use LID to apply the CTP, sometimes I use a command line on the main Lawson server. After the CTP is complete, I have found that Lawson will not run right. I have to run the Stop script, then run the Start script again. That brings everything back up clean.

BTW, you didn't ask about ESP's, so I will volunteer that my experience with ESP's has had the same problem: I have to cycle the entire system after an ESP install. But in the case of the ESP, stuck processes are buried so deep in the OS that only a complete reboot of all servers in the cluster will clean things out. That does the trick.

I suppose that if I had the time to research, I could find the appropriate stuck processes to kill to avoid restart after a CTP or a reboot after an ESP. However I like the cleanliness of a complete restart, as it doesn't leave any nagging question of, Did I really get everything cleaned out?