Reload, Purge Cache, or cycle process

[Todd Lundstedt]

Cloverleaf 5.5 Rev1 on AIX 5.3

I have a TPS proc in a route.  That proc does a table lookup.  When I change the table, what do I need to do to activate that change?

I have always understood that you can “purge cache” to activate a table change.  However, neither purge cache nor reload (of the TPS proc) worked.  The only thing that worked was to cycle the entire process.

Because I am using a tablelookup function from a TCL proc, am I stuck with this as the only way to activate a new table?

Topic

[Jim Kosloskey]

Todd,

Purge Caches should reload the table even if it is only referenced from a Tcl proc (that is never referenced from an Xlate).

That is the way it works on Cloverleaf(R) 5.2.

I certainly hope that has not changed. If it has, I would consider that a bug.

Jim Kosloskey

email: jim.kosloskey@jim-kosloskey.com 29+ years Cloverleaf, 59 years IT - old fart.

[Mark Thompson]

Todd,

The process purge caches is broken in 5.5 (known bug).  You can see it fail in the process log.  Individual protocol threads work OK, but I believe your route TPS runs in the xlate thread, not a protocol thread.  You might be able to get it from a command line.

- Mark Thompson
HealthPartners

[Bob Richardson]

Greetings,

We run Cloverleaf Integrator 5.3 rev3 on AIX5.2 and the purge cache command will work for reloading tables in the xlate thread only.  If you reference tables in Tcl procedures in the inbound/outbound stack you must cycle the process.  That is our experience.

It makes sense in that the table is called via tbllookup and so it most likely is not retained in memory.  It would be advantageous for the engine to keep tables in memory in all contexts.

Enhancement request?

Enjoy.

BobR

[Rob Abbott]

I am fairly sure tbllookup and hcitbllookup cache the table upon first use.  Purge caches on the thread that owns the lookup should clear the cache.

As others have said they have seen purge caches fail – this would be a GUI issue if it’s happening.  When you hit purge caches review your log files to insure that the correct threads are being purged.

Rob Abbott
Cloverleaf Emeritus

[Todd Lundstedt]

In the example listed on the original post, I had purged caches, using the GUI, at the process level, not the protocol thread level.  From what I can tell in the log….

Code:

[cmd :cmd :INFO/0:      Lab_cmd:11/29/2007 14:16:54] Received command: ‘to_fuji_pacs,fr_Mercy_Lab,…more theads than I care to list…,fr_H_softmed purgex’
[cmd :cmd :WARN/0:      Lab_cmd:11/29/2007 14:16:54] Thread is unknown ‘to_fuji_pacs’
[cmd :cmd :WARN/0:      Lab_cmd:11/29/2007 14:16:54] Bad thread-spec ‘to_fuji_pacs’ for cmd ‘purgex’

Oddly, it listed every protocol thread in the site.  I would have expected it to list only the threads in the process, including the xlt_thread.  No xlt_threads were listed.  The “to_fuji_pacs” was not part of that process, so I assume the command failed.  I will try the purge command from the command line next time.

Rob,

Is there an ETA on the fix?  Will it be fixed in a Rev, or will we be waiting for a new version?

You indicated to use the command line and to purge based on the thread that owns the tbllookup.  If my TPS proc is in the pre-xlate position, or perhaps pre-destination position of a raw route, which thread owns the TPS proc; the routing thread, or the xlt_thread?  Can you purge caches on the xlt_thread via the command line?  The command line syntax for the purgex is simply

Code:

hcicmd -p -c ‘. purgex’

To me, that assumes the entire process gets the purge.

Thanks,

[Rob Abbott]

Quote:

Is there an ETA on the fix?  Will it be fixed in a Rev, or will we be waiting for a new version?

This has been fixed in version 5.6, which is due out this month.

Quote:

You indicated to use the command line and to purge based on the thread that owns the tbllookup.  If my TPS proc is in the pre-xlate position, or perhaps pre-destination position of a raw route, which thread owns the TPS proc; the routing thread, or the xlt_thread?  Can you purge caches on the xlt_thread via the command line?  The command line syntax for the purgex is simply

Code:

hcicmd -p -c ‘. purgex’

To me, that assumes the entire process gets the purge.

. (dot) is the command thread.  In early releases I would always include the xlate thread as well to insure it got purged.

I just verified in 5.6 that if you do “. purgex” it will purge the entire process.  Matter of fact, if you purge any thread in the process, all threads in that process get purged.  This behavior also occurs in 5.5 rev1.

I also verified that tbllookup does indeed cache the table and purgex is required to clear the cache.

Rob Abbott
Cloverleaf Emeritus

[Todd Lundstedt]

OK.. I will ensure my Cloverleaf co-workers are aware that we need to purge caches by the command line, and not the GUI.

You indicated it is fixed in 5.6, to be released soon.  Will it be fixed in any Rev update for 5.5?  I just upgraded to 5.5, and I doubt the powers that be will let me do another version upgrade soon.

[Bob Schmid]

I too am interested in 5.5 rev1 offering purge cache fix ????

I have begun the 5.5 / rev1 upgrade from 5.3

[Author]

Replies