New TCL proc, and already-running processes

Todd,

As you have probably already discovered, when adding a Tcl proc which did not exist when the engine was started, one needs to do a reindex (after successful mktclindex) in order for Cloverleaf(R) to have a reference for the Tcl procedure.

The reload is good for when you make a change to an already existing Tcl proc and need to have the new version in memory for Cloverleaf(R).

The purge caches activity does not have any direct relationship to Tcl procs (it does refresh Tables however and if a Tcl proc references a Table…).

Jim Kosloskey

Todd,

We’re on 5.4.1 and beta testing 5.6.

I ran into the same issue during the beta test.

It turns out, as of 5.5, you can no longer make minor changes in a thread and get away with bouncing the thread. You now have to bounce the

process to get changes to take effect.

This is going to mess up some of us “old dogs” who have been using Cloverleaf for years.

-mh

Michael,

Are you indicating that the reindex, reload, and purge caches no longer function in 5.5? We are heading that direction (5.5 that is).

Or are you indicating that if one was used to bouncing threads to activate changes that might no longer be effective and instead bouncing the process is required?

Thanks,

Jim Kosloskey

I am indicating the second.

Bouncing the process is required, bouncing the thread no longer works.

As far as I know reindex, etc. still work.

-mh

Thanks Michael.

I am working in 5.5. Depending on what changes I make I do no have to stop and start the processes only the thread. If I make changes to any non-routed related configuration then stoping and starting the thread works just fine.

Also doing a reload and reinidex works as well.

The things that I have done that do not work without bouncing the process are:

1) Apply enable_all to eo_config in thread definition in NetConfig.

2) Also, changed an outbound ftp thread hostname from alphad to alphaz.

I made the changes while the thread was down, restarted the thread and neither change took effect until I bounced the process. R&D came back and said that started in 5.5.

-mh

Intresting, I have not done those specfic things. I have added tcl procs I failed to add at the start. I have reloaded tcl procs that were updated. I have chnaged the protocol from tcp/ip – tcp/ip-pdl to file. I have not tried just to change a host name I was connecting too.

I have applied the EO-config from the gui and command line and it seems to work ok.

But did you make the eo change in Netconfig?

That’s where I’m having the issue.

I just got off the phone with Rob Abbott and I may have bad info.

My issue may be a bug in 5.6 instead of a design feature of 5.5.

This may not be a problem with 5.5.

I’ll let you know what I find.

The only thing I am aware of that you may have to cycle the process for is if you are using MQ.  The MQ parameters are maintained under the process.  Of course it has been that way for a while.  The rest is still the same as always; cycle SMAT, reload, etc.

Michael:  I hope you clarify before you scare everyone off of 5.6  😆

So far, I have found 5.6 to be VERY Cool!

It is not my intention to scare anyone, Halloween is over.  😈

My bad for opening my mouth.  😳

All,

reload, reindex, purge caches work fine in 5.6.  

Mike did find an issue in the 5.6 beta where protocol threads were not picking up changes in NetConfig when they were bounced.  This will be fixed before a production release.

Thanks