Some of my regions are rocking some longer uptimes then previously possible and doing great despite idle scripts using more time than seems reasonable.

[Ardy Lay]

Just two examples;

Simulation of region <redacted> has been running 21 days 0 hours and 36 minutes.  (Not a lot of traffic on this one.)

Simulation of region <redacted> has been running 16 days 19 hours and 54 minutes.  (5 to 7 parties a week, although the crowds have been light recently.)

Both are running Second Life Server 2020-09-11T22:25:15.548903 in the co-location facility.

I am looking forward to simulator hosts getting 'up-lifted' because the recent software updates I attribute to facilitate running Second Life services on AWS have improved simulation longevity significantly.  I do, however, still see much 'time' being used by scripts that are just waiting for an event to be triggered or otherwise not doing much.  Since this awkward consumption of scheduled resources arose, I have seen several devices with active scripts get less than normal run-time and behave abnormally as a direct result.

I would like to cheer you on, Lindens and Residents alike, and encourage you to update JIRA issues about this script time issue once the regions are all up-lifted and the dust and debris is cleared away.

[Ardy Lay]

Simulation of region <redacted> has been running 26 days 23 hours and 35 minutes.

And still doing fine.  The other got "up-lifted".

Edited November 9, 2020 by Ardy Lay

[Lucia Nightfire]

On 11/2/2020 at 8:41 PM, Ardy Lay said:

I do, however, still see much 'time' being used by scripts that are just waiting for an event to be triggered or otherwise not doing much.  Since this awkward consumption of scheduled resources arose, I have seen several devices with active scripts get less than normal run-time and behave abnormally as a direct result.

Can you be specific as to how said active scripts are behaving abnormal?

The increase in idle script timing was something I brought up in the server meeting on Oct 27 and this jira was filed.

[Ardy Lay]

Oh, just that scripts waiting for events appear to be taking more time per frame than they did several years ago when we built the venue.  As a result, interactive objects are no longer  interactive.  Now when we touch an object we have to wait some apparently random period of time to get a reaction.  I don't remember exactly when this change happened but it wasn't recent.  Animats, Qie Niangao and others were discussing it here at the time.  I think the big question was when did this start happening.  I think the group narrowed it down to a 2 or 3 month time range but that was years ago now.

[Lucia Nightfire]

7 hours ago, Ardy Lay said:

Oh, just that scripts waiting for events appear to be taking more time per frame than they did several years ago when we built the venue.  As a result, interactive objects are no longer  interactive.  Now when we touch an object we have to wait some apparently random period of time to get a reaction.  I don't remember exactly when this change happened but it wasn't recent.  Animats, Qie Niangao and others were discussing it here at the time.  I think the big question was when did this start happening.  I think the group narrowed it down to a 2 or 3 month time range but that was years ago now.

Is the Scripts Run rate for the AWS region in question below 100%?

If so, was the Scripts Run rate higher before the move to AWS?

Edited November 9, 2020 by Lucia Nightfire

[Ardy Lay]

This is fun. My quiet home region in the datacenter: Simulation of region <redacted> has been running 32 days 0 hours and 0 minutes.  

This is not so fun.  I have been watching the % scripts run on regions moved to AWS.  It's hard to really say there is a large difference.  Seems like running idle scripts might be taking the same time as before the move or they might be taking a little more.  I haven't built a test lab.  😉  Also seems that scripts that do a lot get it done a little quicker then return to idle and sandbag the rest of the time.  This makes me sad.  I used to be able to have 8,000 idle scripts hanging around waiting for an event without really hurting the 1,200 or so active scripts in the venue.  For a couple of years or so those numbers are cut in half and moving to AWS hasn't helped.  As a result, I have been buying out other land owners in the region then leaving the land empty.

[GamesDeluxe SLSGO]

On 11/14/2020 at 12:58 AM, Ardy Lay said:

This is fun. My quiet home region in the datacenter: Simulation of region <redacted> has been running 32 days 0 hours and 0 minutes.  

This is not so fun.  I have been watching the % scripts run on regions moved to AWS.  It's hard to really say there is a large difference.  Seems like running idle scripts might be taking the same time as before the move or they might be taking a little more.  I haven't built a test lab.  😉  Also seems that scripts that do a lot get it done a little quicker then return to idle and sandbag the rest of the time.  This makes me sad.  I used to be able to have 8,000 idle scripts hanging around waiting for an event without really hurting the 1,200 or so active scripts in the venue.  For a couple of years or so those numbers are cut in half and moving to AWS hasn't helped.  As a result, I have been buying out other land owners in the region then leaving the land empty.

You can test it yourself, while you can still find non-AWS regions, and see how they perform comparatively. We're looking at 15%/20% higher Script Times of the default script. 🙃

https://jira.secondlife.com/browse/BUG-229617?focusedCommentId=639234&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-639234

This simple test can be quickly performed with Firestorm.

That comment is from a "duplicate" JIRA, this is the JIRA that Lucia started and stands as the main one. https://jira.secondlife.com/browse/BUG-229611
 

I was told that all remaining regions are going to be moved to AWS today (hope not, I still have 1 'good' one), so test fast 😋