[Sophie Chege]

Thank you Lonnie. I will do the same.

[Sophie Chege]

Thank you Jeff

[Sophie Chege]

Peter-

Some performance issues with 6.2 affecting AIX customers are addressed in 6.2.1 (See Rob’s note about 6.2 and AIX). All of the issues are supposedly fixed in 6.2.2 which was slated to be released in May.

[Sophie Chege]

Sorry for the late response:

We are stable..ish, I say this only because we have now figured out how to quickly resolve some of the issues with 6.2 that we did not see with the previous version without any calls to Infor

Infor has been very helpful and has worked with us, and is still working with us.

We have not applied any patches as of yet (still on 6.2.0). (We are a little apprehensive)

[Sophie Chege]

We document our tcl’s  really well to ensure that if one of us gets hit by a bus, the rest of us can follow their code but for interfaces, we have a job that tars up the sites in entirety and they are backed up in our disaster recovery locations, as well as having an active:passive structure in case the active box goes down.

[Sophie Chege]

No worries James, ask away. Maybe someone can see something that we are missing.

1. Our hardware is 100% dedicated to Prod.

2. Our volume is double what you have. I should also mention 1 site is master. so technically 6 sites.We average 8 processes per site.

[Sophie Chege]

We are on AIX 7.1 TL3. Our box has 24 CPU’s and 192GB RAM. Here is what we have noted. Everything has been reported to Infor and I do have to say that Infor development has been actively working on our issues.

1. Site capacity has greatly decreased. We were running CIS 5.8 on a less powerful box (AIX 6.1 4CPU, 16GB RAM). Our 6.2 sites seemed to crash with the same amount of threads/processes and volume as what 5.8 could handle without any issues. We had to break down our sites from 5 sites to 7 sites. we had to split some processes too as they did not seem to be able to handle the volume that they handled in 5.8

2. Hostserver constantly went down after we went live- probably because of #1, and who knows what else.

3. Depending on location of tcl procs, processing of messages is affected. We have routing threads, may not be the best set up ever but no issues in 5.8. In 6.2, messages seem to get stuck at different states. Mostly they were stuck in state 5 in the recovery database. This would also cause # 2, when they got to be too many. We resolved this by moving our tcl procs from the routing threads to the actual receiving thread …there’s prob a better way to explain this 🙁

4. Processes get hung and it takes a long time to stop and start. (I’ve waited 10 mins). Nothing in logs to indicate the cause.

4b) Inbound threads  losing connection but the display remains the same- all green and up,hciconnstatus shows all up but sending system showed no connection, so we had to set up alerts esp for our ops  that monitors 24/7. A simple bounce fixes this (could be argued as networking but we did not see in 5.8 )

Some minor issues:

1. cannot resend from error database if threads are in 2 different processes (bug)

2. process start times appears to be later than thread start time (I’ve posted this one here before.)

3. We were going to enable GM but I’m wary since everytime I call support they ask if we have it enabled. We went live with GM but we turned it off as per Infor’s recommendation.

4. When you open Smatdb, IMHO, the default should be all disabled and not everything enabled to search

Other than the occasional headaches,  we love the new features that 6.2 has to offer :)..oh and Infor implementation team helped with the upgrade, so we did not go rogue with the install process.

[Sophie Chege]

Be aware that 6.2 may not be as stable as it is. We went from 5.8 to 6.2 in March and we were more stable in 5.8. We have had nothing but issues in 6.2. We stayed on Unix

[Sophie Chege]

Mike-

We have the same issue with the exception of the ulimit settings. Our settings are all in alignment with the proposed changes. Did this change resolve this issue for you?

[Sophie Chege]

Thanks Robert-

We ended up using CAA_WS  and route replies with tcl to add the wrapper to the msg going out and tcl to remove the wrapper to the message coming back in.

[Sophie Chege]

Thank you!!. I was watching out for this reply as well.

[Author]

Replies