execution_contexts table --> number of records??

tford

I believe that growth in the execution_contexts table over the weekend causes our server to come to a screeching halt yesterday.

I'm trying to see how many records others report in this table. Could you tale a moment and look under the hood at your database and report the number of records in this table below?

Thanks!

kklosson

These are hung processes (may not be an accurate term). On my admin console, I present all active processes to keep an eye on things. You can launch that from the admin menu. I had some processes that needed to be optimized. I didn't know what Vlad meant be that but I have a better idea now. Processes spawned by users, in my view, should not last long or take significant resources. You can safely delete these with the AIM server offline. I would not expect to see many items in there. But each of them, AFAIK, constitutes either a running or hung process. If you look at the active processes, you can see the state. Waiting is not good, probably hung.

In my case, I found that a process that launches a form for editing is maybe not optimal. Others may be more knowledgeable...

tford

Hi KK,

On my admin console, I present all active processes to keep an eye on things. You can launch that from the admin menu. I had some processes that needed to be optimized.

I'm running multiple business spaces, so it's not real practical to monitor this at the business space level.

These are hung processes (may not be an accurate term).

This was post by support:

There was another forum process that said processes with "allow user to cancel the process" unchecked are candidates for being left on this list.

My issue:
During the past weekend, the execution_contexts table grew substantially:
-- 180 records at 10 pm on March 11th
-- 3,744 records at noon on March 14th

As a result, the system came to a screeching halt Monday morning until I could revert to a backup. If I had identified execution_contexts as the issue, I could have probably just deleted the records in that table & the system would have been fine.

kklosson

For me, a hung process is a process listed in the active processes list that lives well beyond any reasonable timeout. Like 12 hours or something.

tford

For me, a hung process is a process listed in the active processes list that lives well beyond any reasonable timeout. Like 12 hours or something.

I get where you're coming from. Most of the processes I see in the execution_contexts table make no sense anyway. The only ones I understand are put while a process is running a major update (for example reading through instances of a BO IN BATCHES OF 1 ...)