Who can I get a noticication that one of our Runtime server is "dead"

S

SERBEM

Member
Hello there,
We sometimes have a issue that one of our runtime servers going "dead", for example last time there was an unexpected pop-up window infront and the robot just thought it was in the right application as it could find the application main window - so our runtime server just was dead from Friday night until Monday morning :-(. We dont yet have a 24/7 service as we only a very tiny department. Is there some way we can get a notification that the server hasn't done anything for quiet a while? We contacted the server gays and they cant help us.
 
V

VJR

Well-Known Member
Hi SERBEM,

At first take a look at Alerts and see if it works in your scenario.
http://rpaforum.net/threads/alert-stage.413/
Read all the posts in the above thread as there is key information like Alerts require subscription per license, etc.


so our runtime server just was dead from Friday night until Monday morning
Whether Alerts (or any other method) will work or not will depend on things like what happens when the unwanted popup comes up. Does it get stuck there infinitely on any particular stage?..if yes which one? Or if an Exception Handling is present in the diagram does it go into that? Does the process show up as Failed? Or nothing happens at all? What does the last line in the log indicate?

Here is a good video on the Alert stage.

You can do some research & analysis on Alerts and then post back here and then we can think of any other alternative.

While you do the research on the above, I'm wondering what is the notification you are expecting since you made a note that the team is not 24x7. So ideally nobody would be able to see any Alerts on the machine. So how would you like a notification? Do your team members have access to email so that emails are sent whenever a process is stuck or failed? Post back with what all options you have in hand and the inputs to the above queries.
 
S

SERBEM

Member
Hi VJR,


Thanks for your reply, we will have a look into this Alerts function, we ATM only use Alert with termination, by message to the RPA team and an e-mail to the department owning the process.

Our RPA team getting termination messages 24/7 but we only prior critical processes out site working hours, but if we can get a mgs that one server is "dead" there should be a chance some of us can solve that, also out site working hours.

Our problem have been that once in a while some of our process dies in the middle of a process, like BP think it’s working but nothing happens. It especially in one of our old systems, where you don’t always know what happens - it’s strange to complain ;-). We don’t have that amount of problems in our other applications.
 
N

nickbell123

Member
SERBEM said:
Hi VJR,


Thanks for your reply, we will have a look into this Alerts function, we ATM only use Alert with termination, by message to the RPA team and an e-mail to the department owning the process.

Our RPA team getting termination messages 24/7 but we only prior critical processes out site working hours, but if we can get a mgs that one server is "dead" there should be a chance some of us can solve that, also out site working hours.

Our problem have been that once in a while some of our process dies in the middle of a process, like BP think it’s working but nothing happens. It especially in one of our old systems, where you don’t always know what happens - it’s strange to complain ;-). We don’t have that amount of problems in our other applications.
Click to expand...
A way my team built to catch these sorts of issues (stuck resources, failed schedules, over running schedules) is to maintain a list of your processes and store a variable to define how often they should "check-in". Essentially a heartbeat process.

In our case we have a list of processes on SharePoint, each Blue Prism process calls a sub process in the case working loop to check this list. If the time that has elapsed since they last updated the list (or "checked in") exceeds a certain limit, they update the list again. If they fail to update the list, the support team can see all the processes and resources which are "stuck". There will be many ways to implement such a solution - the important thing is to have a way for the Blue Prism process to talk to your list/end point and a way to monitor/alert your team if the link is broken.

See Keepalive for a generic description.
 
V

VJR

Well-Known Member
SERBEM said:
Our problem have been that once in a while some of our process dies in the middle of a process, like BP think it’s working but nothing happens. It especially in one of our old systems, where you don’t always know what happens - it’s strange to complain ;-). We don’t have that amount of problems in our other applications.
Click to expand...
Hi SERBEM,

If this is happening as a one-off case then you can have a logging feature in this particular process.
A simple writing of 1 line to a text file with a DateTimeStamp and the log message. This can be done using a simple BP action (Append to Text File).
eg;
29/09/18 07:00:00 - Process Started...
29/09/18 07:01:00 - Get Worksheet as Collection executed...
29/09/18 07:02:00 - Looping collection...

These are just examples and you can customise them as High/Medium/Low type of logging messages which can be parameters to the new Write2LogFile action where you can set the depth level of the logging messages.

Have this log file created in a designated folder - C:\BPLogFiles\ProcessName\
Write a simple vbscript to read the DateTimeStamp of only the last log message.
If there is too much of a difference in the DateTimeStamp from the log file and the current system datetime then it will send an automatic email to all the designated folks indicating something is wrong.
You can have this vbscript file set to run in the Windows Scheduler.
 
You must log in or register to reply here.
Top