recover_33 procs: What does one do after so many retires

[Keith McLeod]

What is the best strategy for a thread that keeps retrying a message and the receiving system won’t acknowledge?  I have established a counter to allow me to do something based on the number of retries.  For instance, I can send out an alert every so many retries as an escalation.  At some point I would think to shut down the thread to force a response through normal operations channels.  What has been done and has been acceptable by the receiving system administrators?  I would like to hear some ideas.  Thanks…

Topic

[Bryan Dort]

I have multiple levels of alerts setup for different threads…depending on their activity.  

For instance, on one thread, I have an alert that checks if there are 5 messages backed up in the outbound queue for over 60 minutes.  If so, then I run a windows .cmd to bounce the thread.  If over 90 minutes, then I get paged.  Most of the time, bouncing the thread cleans things up.  On more patient-care critical threads, I don’t wait as long.  

Some systems can’t handle the bouncing as well as others.  So I don’t even bother to bounce them since I’d have to restart the interfaces on the remote servers anyway if they lost connection.  You could bounce the thread and page the system administrator of the remote system to restart their interface.

[Author]

Replies