[Vince Angulo]

Gina,

We have a similar site configuration to yours; the only difference is that the analyst that does web service connections uses a separate site for those processes.

We’re a medium-sized pediatric integrated delivery system with 1 acute care hospital, 2 ASCs, 3 urgent care and 6 diagnostic centers, and 20 gen peds physician practices.

We have connections for 65 end-to-end interfaces, using about 100 of our unlimited thread license. Six total sites and their general contents:

• Hospital ADT, BAR, and DFT
• Physician Office Reg/SIU
• Orders (Nursing orders to bolt-on downstreams – audiology, nutrition, EEG, and the like.)
• Results (Reference labs, general transcription, PHR eLR)
• Ancillary clinicals (Mainly Pharmacy, Rad-PACS-Transcription, Blood Bank)
• Web Services (Continuity of care)
• Master (Reusable tps and xltp scripts)

We try to be conscious of Infor best practices for things like # of threads per site…so, that drives our set up.

Current staff is 2 -1/2 FTEs for interfaces.

I’ll be following this topic with interest as we are transitioning from Epic Physician Practice Reg/Cerner EMR to an “All Epic”, “Epic First” vision. Dropping Hospital ADT and billing in favor of Epic HAR as well as adopting any Epic integrated solution that can replace our bolt-on solutions.

We are also simultaneously changing from locally hosted Cloverleaf on AIX Power 10 to Cloverleaf “SaaS in the cloud”.

We kick off this spring with a projected Epic ‘big bang’ go-live in Fall 2027. For this project, we will be increasing to 3 FTEs + 2 temp consultants. The temp consultants will support existing interfaces, with the original 3 focusing on building new interfaces to meet the project vision.  When we’re fully on Epic the temps will be let go.

[Vince Angulo]

No experience with the first issue, but we used to see the second issue from time to time.  Haven’t seen it in years, but here’s what we have in our troubleshooting wiki:

Entire site:

The hcimonitord is hung.

Use ps -ef | grep hcimonitord to get the pid, then kill -9 <pid>.  Restart the process with hcisitectl -s m.

One process:

The hcienginewatch for the process is not running so threads in that process show as ‘dead’ and will not start.  Threads are probably running, attempt to verify on target system.

Use ps -ef | grep <process name>.  It should return two results the pid for hciengine and the pid for hcienginewatch.

If there’s no hcienginewatch, the Network Monitor becomes unresponsive because it thinks the process is stopped, and won’t start because hciengine is still already running.

Resolution is to kill -9 <pid> for the hciengine process, then restart the process from the left side of the Network Monitor as usual.

======================================

A site init (db clean up) should be scheduled when possible

[Vince Angulo]

I have this happen with our Cerner feeds — the Cerner segment organization doesn’t always follow the groupings that Cloverleaf variants default to.

I end up running the input message through the HL7 testing tool to verify which group Cloverleaf is pulling segments into.  In particular segments like NTE and OBX which may appear in a number of different groups.

[Vince Angulo]

This was long time ago (v5.4), and I don’t have the detailed results anymore, but here’s my recollection.

What I did was create a few file to file interfaces.  One with a tps proc to filter on ADT types, one using an xlate, and one using a reg exp in the routing.  I believe I was filtering out merge and delete transactions (A34/35 and A29).

Created a file of 20K messages, ran them through each interface. This was on our test server with almost no other interfaces competing for resources. (AIX).

If I recall correctly, the tps proc was faster by several percentage points.  Routing edged out xlate, but only modestly so.

[Vince Angulo]

Thank you Todd!

[Vince Angulo]

sorry the copy and paste will not put in the entire join statement, and the edit won’t let me type them in either.

If anyone is familiar with this please e-mail me at vincent.angulo@chkd.org and I can provide the entire code clause.

 

• This reply was modified 4 years ago by Vince Angulo.
• This reply was modified 4 years ago by Vince Angulo.

[Vince Angulo]

In the VRL tool, go to the Options menu, Global and you can define delimiters in that dialog box. They’re called Global, but they only apply to current VRL.

[Vince Angulo]

A brute force way to do this might be to concat a string of carets to the last value.

[Vince Angulo]

We don’t use this, but I believe the Secure Messenger add-on is required leverage this type of connection.

[Vince Angulo]

Thanks Jim, I was pretty sure that would be the case.

For me it’s a maintenance convenience.  Since I haven’t run into the HL7 version with different numbers (yet), I can find, say, the patient name, by searching 00108 and find everywhere the translate touches that field.  This is esp. useful as we have one developer with a quirky organization style…

Over time, I’ve become really familiar with the numbers and it’s quicker than trying to search a field using the # notation — I don’t have the PID(0).#5 on the tip of my brain — if that makes sense…I guess that’s my quirk 🙂

[Vince Angulo]

I haven’t found anything useful and had to contract with Infor for implementation assistance.

The resource I was assigned would not help me understand how to use the tools and insisted on doing the work himself at a pace that I could not keep up.

It’s not helpful, but was my experience…

 

• This reply was modified 4 years, 7 months ago by Vince Angulo.

[Vince Angulo]

We cannot browse above above the site directory.

But we can type in a full path in the Outbound Directory to access our attached storage.

[Vince Angulo]

if “filter” is the value of the inbound  OBX#3, then it should work.  If “filter” is the result of the table output, then the IF criteria needs to use the outbound OBX#3 value, i.e., it should be preceded by a tilde/~.

[Vince Angulo]

We also use SMAT as the source of truth for this.

We’ve tried saving the ACKs to SMAT as well, but have had issues with it working consistently — probably something wrong in my build, we’re still fine tuning it.

It would be helpful if I could tell when the process writes the message to SMAT.

I’m guessing it’s sometime after State 14, but can’t see anything in an ‘enable_all’ process log that says the message is being written to SMAT.  Does anyone know where this happens?

[Vince Angulo]

As long as the $new_name variable does not change with every transaction, if you check “Append data” in the file protocol Outbound Properties, it should work.

[Author]

Replies