Hi Folks, This is regarding engine checkpoints. It is my understanding, tha

uril over 9 years ago

Hi Folks,
This is regarding engine checkpoints. It is my understanding, that during checkpoints the gateways (which affected by the process) are temporary not available for processing. Now, if we have let's say a significant size and many gateways for process-exec and others, it basically make the system "frozen" during checkpointing procedure. And, if checkpoint is scheduled, we often see "warn checkpointing is needed" instead of "OK" from checkengine result.
Now, what if we have a scheduled script, which will look for those "checkpointing needed" messages and initiate the checkpoint just for that particular gateway, then wait and proceed to the rest of the list, and continue to check all the gateways during its run? In that case (IMO), only processes need to be check pointing will be fired, and rest will be available for processing.
So, Appian Gurus, is it something worth to consider or it really doesn't matter from the system point of view?

Att...

run_warn_checkpoint.sh

Discussion posts and replies are publicly visible

0 uril over 9 years ago

...aching a simple script, which works for me for analytic and process gateways. I wonder what you think.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 nathan.schmitz
Certified Associate Developer
over 9 years ago

The checkpoints are typically not a problem if you are following other best practices. Checkpoints become problematic if your kdbs are too big.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Chris over 9 years ago

We don't often see issues during checkpointing and our KDBs are in the range of ~18 GB each. Only concern for us is scheduling nightly processes around these times if they are heavy on database transactions.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Tom Ryan
Appian Employee
over 9 years ago

You can most likely remove the complexity of monitoring for checkpointing needed messages and run a scheduled checkpoint as required, staggering the checkpoint for each engine to distribute load more evenly. More info in the docs, in case you haven't seen it yet:

forum.appian.com/.../Configuring_Application_Checkpointing.html
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 uril over 9 years ago

Hello colleagues Thanks for your response. We have close to 20 Gb Kdb files (mostly exec), and growing. We can schedule it of course, like twice a day for each gateway. Right now it is not even scheduled, so Appian itself manage it (maybe that is why I see "Warn Checkpointing recommended" for all my Process-exec.
My point was even if scheduled, if we have increasing wave of transaction it still might come up.
Based on the answers so far, it seems make sense to schedule (as many times a day as needed for all or for each gateway, and don't worry those messages even if they come up from diagnostic/checkengine report.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Tom Ryan
Appian Employee
over 9 years ago

It may also be helpful to review your policy for archiving/deleting old processes that are no longer needed. This could help to prevent the exec engines growing so large:

forum.appian.com/.../Data_Maintenance.html
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel