Jump to content

Deploy Plan for the Week of 2021-12-06


Mazidox Linden
 Share

You are about to reply to a thread that has been inactive for 830 days.

Please take a moment to consider if this thread is worth bumping.

Recommended Posts

On 12/8/2021 at 3:42 AM, Mazidox Linden said:

Tuesday: Weekly rolling restart

@Mazidox Linden

After all the bashing LL have been taking recently I thought you might like a bit of positive feedback.

Just before the big crash today, one of the mainland parcels I have some stuff at got restarted. It was a surprise as it's main-channel not RC, but when it came back up and I got de-crashed I found it was now running all the objects there much more smoothly, and the scripts run is up at 85%. This is now the second such mainland parcel where I've now got things rolling around as smoothly as they used to do back in 2018 with less worrying stats, so the "configuration change" you mentioned back in November as possibly bringing some performance improvements most definitely has. My thanks to all of the people who conspired to bring this about.

Link to comment
Share on other sites

Maybe you're next on the configuration change. There were several weeks gap between my first mainland parcel getting improved and today's one.

Have you tried some of my region test scripts to see what results they give?

 

( the reply was for for Aishagain , Henri sneaked in as I was clicking the post button :)

Edited by Profaitchikenz Haiku
Link to comment
Share on other sites

@Henri Beauchamp what is 566713, is that what was rolled to some of the RCs?  Maz makes no mention of new code for this week, but then there is nothing posted officially for this week at all.

@Profaitchikenz Haiku:  Monty Linden was quite insistent that we'd been moved to the new config 2 weeks ago, though I have nothing to show for it.

Edited by Aishagain
Link to comment
Share on other sites

7 hours ago, Aishagain said:

@Henri Beauchamp what is 566713, is that what was rolled to some of the RCs?  Maz makes no mention of new code for this week, but then there is nothing posted officially for this week at all.

It just got rolled to my home region (and a few others around mine in Zindra), which was so far running Second Life Server 2021-11-11.565737... Perhaps in reply to the concern I raised during the last Simulator User Group meeting about failed event polls in neighbouring regions, but I am still seeing these... Or perhaps was it rolled to regions that failed the first restart after the rolling release (mine and a few others around it were affected and took 3 hours before they were finally restarted).

Edited by Henri Beauchamp
Link to comment
Share on other sites

The general kerfuffle on Wednesday, which produced an enormous collapse in concurrency, was attributed by Linden Lab, in a status page, to internet problems. Other public news media reports an AWS failure, at the right time and in the right geographic region.

Inara Pey, in her report on the SUG meeting. gave a version number for Tuesday's MC release, 566406, described as necessary for the change to IM/email options, but while something new was deployed the RC channel, no version number was reported.

This should be the last server deployment for this year

The SUG meeting nest week is apparently intended to be a Solstice Party, but even if it is transferred to an alcohol production facility, I have doubts about Linden Lab planning. (NB: I am British...)

  • Like 1
Link to comment
Share on other sites

10 hours ago, Aishagain said:

Monty Linden was quite insistent that we'd been moved to the new config 2 weeks ago, though I have nothing to show for it.

What about the neighbouring regions to yours? In trying to make sense of why the final mainland parcel I was watching came good yesterday, I was wondering if having adjacent regions on different configurations would be sub-optimal and they'd only show improvements when all the boundary regions had arrived at the new configuration.

The other thing of course is what performance degradation are you seeing? I have realised now why Monty was a little dismissive of people who studied the Scripts Run %. It's easy to believe it describes a continuum, but I now think of it more as discrete bands of performance. 100% is excellent, but all the way from 85%  down to 60% I am seeing much the same performance. Until yesterday the 35% region was showing seconds lost when avatars entered or left the region but now it's not showing any such issues.

On it's own the figure doesn't tell you a lot. Run some keyframed motion objects or movetotarget ones, or have a script using a 1 second timer trying to do something precise and there were distinct glitches prior to yesterday's restart.

I'll have to monitor this for a few days, but the test script I had carefully designed to try show the knock-on effect to time drift when avatars entered a region seems redundant, because there is now no such measurable effect. Whatever they've done certainly seems to have made a significant change in that particular respect, also I am seeing far less moving_end timeouts, so there has to have been an improvement n how a moving object is tracked against criteria and the moving_end event raised.

Edited by Profaitchikenz Haiku
Link to comment
Share on other sites

@Profaitchikenz Haiku: Woods is a standalone region, though obviously there will be other regions on the simulator but I have no idea as to which or how many.

With the script-run figure of 35-45%, performance is only mildly affected but occasional delays of a frame or two with some scripted functions, noticeable with a couple of animesh creatures, are seen.  I don't have them rezzed permanently so I don't  have data on long-term activity.

 

Edited by Aishagain
Link to comment
Share on other sites

@Henri Beauchamp thankyou.

& @Linden Lab Why is there STILL no official information on what was rolled on Wednesday and might, by inference, be rolled to Main Server regions next Tuesday?

Mazidox' incapacity is NO excuse for the lack on communication.

Edited by Aishagain
Link to comment
Share on other sites

@Linden LabAnd we are now on the eve of yet another round of rolling restarts with no information from LL as to what we are to expect. 

Have you abandoned any pretence even of informing your users?

ETA Monday 3pm PDT:

I just looked at the Grid Status Page and all mention of rolling restarts for both Main Server and the RC Channels, which were there over the weekend has vanished.  So I must assume that there will be no new code rolled, since the routine restarts rarely warrant a post on the GSP until they are happening.  Maybe it's just me but this seems a bit haphazard.

Edited by Aishagain
Link to comment
Share on other sites

You are about to reply to a thread that has been inactive for 830 days.

Please take a moment to consider if this thread is worth bumping.

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
 Share

×
×
  • Create New...