[Rob Lindsey]

So we have built a singular script to shutdown and to startup Cloverleaf and everything else that we do on our engine systems.  It is a manual process because you just never know when the system has to do work before anything starts up.  We have 1 system that has 25 sites with 294 processes and 4,000+ threads.  Our shutdown script takes about 15 minutes to shutdown everything.  Each process in each site has the command hcienginestop run against it so it does the “normal” stop and handles the killing if necessary.

[Rob Lindsey]

Back in the day we did this type of thing where there was a call to a database server on the network and we did a single call and used a Cached db connection that was opened.  So in the “start” section of the tclprocs that were using db calls, we would use a Global variable for the connection to the DB.  So if the global variable did not have anything in it… then we would do a DB open call and set the global variable and then it would be available to all of the tclprocs within that process.  Of course, you would have to put a disconnect in the shutdown section of the tclprocs so that the DB connections are closed.

[Rob Lindsey]

Hello Jim,

Long time since you and I have interacted.  We will be spinning up these in 6 weeks.  I can let you know how that goes as we plan on getting these up and running in production this year.

Rob

[Rob Lindsey]

Yes it is possible to do but you have to be very wary of many things esp LOCKs and if there are changes from version to version.

I have done this type of programming in our environment.  We do this because of our design and how many interfaces we have (over 7000) so it is easier to do Autostart and a few other things (port, host, bindings, etc).

Rob

[Rob Lindsey]

Thanks for the hard work.  This is a great resource of info and people.

Rob Lindsey

[Rob Lindsey]

Rick,

Not sure where you are on this but are you basically trying to run a program to look at the Error DB entries on a schedule?

If that is yes, it is possible to do. We have a Shell script and TCL that run out of CRON (I’m old school) that look at every Error DB ( elog.elogdb ) on each one of our CL systems.

This is the select command that we use in our TCL
set sqlcmd “SELECT SourceConn, MidNum, EdbState, ErrorString, UserData, MessageContent, TimeOut FROM elog_msgs where datetime(TimeOut/1000, ‘unixepoch’) >= datetime(‘$timeStart’, ‘unixepoch’);”

Rob

[Rob Lindsey]

We have developed some web pages that use LDAP for checking to see if someone belongs to a group and we use TCL.
<pre>package require ldap
# Connect, bind, add a new object, modify it in various ways
set handle [ldap::connect <ldap server> <ldap port number>]
set x [ldap::search $handle “<breakdown>” “(userPrincipalName=$username)” {memberOf} ]
ldap::unbind $handle
ldap::disconnect $handle</pre>
The above code is very abbreviated but might give you some ideas.

[Rob Lindsey]

Thanks for this.  Web service stuff is still a bit of a black box to my brain.  I have found out more info.  Web service inbound will need to receive orders.  It will have to ACK those orders and then send them on.  The outbound results will be two of them. 1) client calls web service to pick up results so I know that I will have to call some type of DB to get the results for the client that logs in.  I need more details on how we will be getting that client login and password to the actual result. 2) a web service push – I will need a lot more information this one to design it properly.

As Johnny-5 would say…. “Need more input”

Rob

[Rob Lindsey]

Well I was hoping that someone figured out a way to use the recovery DB in that shape form or fashion.

[Rob Lindsey]

I have seen the same issue in 20.1 and 20.1.2. From what I can tell it is trying to revert the changes since you made a change but did not apply it.

Rob

[Rob Lindsey]

Jeff,

We have found that SMATDB is a useful way to store data for inbound and outbound saves on threads. There are WAL and SHM files that might have the data but doing a SMATDB search via the IDE/GUI should pull that information in.  I have not found where the data messages are missing.  The hardest part is matching messages when outside of the IDE/GUI.  It is faster and smaller footprint than the IDX files.  I have not found in the IDE/GUI where it does not put them into the proper order when I sort them by time when I do it.  I usually use our tools though.

Rob

[Rob Lindsey]

This is all speculation as this new thread (web service) is going to replace an existing service and I don’t know how that one works as of yet.  I am hoping that I could use the standard recovery DB and just have the results waiting there and pick them out by certain values in the MSH segment.  Of course, I don’t know how the current one works and wanted to know how others have setup their Web service threads that listen for connection and send information out after the connection.  It does sounds like calling another service to get the results for the specific client might be the best option.

I don’t like it when I don’t know what I don’t know.

Rob

[Rob Lindsey]

Well this is for results where many clients would be connecting.  I was hoping to use the recovery DB within the site that the thread would be running in.  I do not know exactly where the results are coming from as of yet but I was making the assumption that I would have them coming in on an inbound thread that would send them to a outbound thread that would listen for connections on the URL.

[Rob Lindsey]

I have not seen this before.

I assume you are using the SMATDB IDE/GUI tool.

I have my own tool that I wrote to do that type of thing.

Rob

[Rob Lindsey]

This is how we do it with our 20.1 environment using RHEL

18,48 * * * * source /home/hci/.profile ; /hci/cloverleaf/scripts/export_interface_info_tables_to_web_page_v3.tcl >> /info/logs/export_interface_info_tables_to_web_page_v3_$(date +\%Y\%m\%d).log 2>&1

The source command pulls in the profile which makes sure that all of the Libraries are set correctly for our environment.

Rob

[Author]

Replies