This week we were expecting, and received, another slew of rolling restarts as part of the final push (or so Oz Linden had said) to get SL totally into the cloud.
The Rolls on Monday and Wednesday appeared to go to plan, but then it became clear that there was a problem updating EEP settings on all regions. It seemed that the second update to the servers had introduced either a bug or a regression and EEP function was corrupted.
The solution was a set of back-end and server-side changes necessitating a further roll to all sims on Friday.
All well and good except….. the rolls did not occur when they were supposed to though nearly all sims were rolled in due course. I say nearly all because my home region (and several others at the least since there must have been other regions on that server) did not get the updated software version and as a result EEP still could not be customised. However that was not all! The “back-end” changes had caused a communications glitch for scripts that needed to communicate with external servers. That issue was quickly and efficiently dealt with and the weekend began.
However , when I contacted SL support regarding the missed regions, I was astonished to find that the support operative had no idea what to do about it and eventually told me that I (and the other region owners, I assume) would have to wait for the next round of rolling restarts before my region could be put on the fixed server version! It not transpires that LL, at the weekend at least, cannot effect rolls and thus correct errors!
I was offered a region restart, but since that would not change the server version, it was irrelevant. So as of now I await developments, but I suspect this is not going to be performed any time very soon.
A shame, it had all been going so smoothly until this week and now the uplift, as far as I am concerned, has crashed and burned.
One aspect of this bothers me: just how much control does LL actually have on these uplifted AWS servers and are we now dependant upon AWS techs to perform rolls? That seens to be an unreasonable penalty to pay for the presumable saving in running costs.
ETA Jan12th 20201: Just to clarify all the above, I was told later that week by Mazidox Linden that indeed some regions had been deliberately NOT rolled to give them a comparison for the new code rolled to most regions, though there was no explanation WHY they decided to do such a thing. Subsequently it became clear to users that all was not quite as it seemed and LL were not inclined to explain their actions.
Over the period between then and now a few more restarts have occurred, the last one on Friday last between 10am and 2pm SLT (brilliant timing, one of the peak periods for events for European users) was simply to re-roll identical code but since the same code could not be rolled twice (why?) it was renumbered. No explanation has been forthcoming as to why such an action was needed and no performance issues seem to have come as a result, but while LL have trumpeted the end of the uplift, they tell us that some systems are still not working due to code chages being necessary. One casualty of this is the World Map, which still is non-functional to a large extent. This has made sailing long distances a greater lottery than ever, though I have been told that the Gulf of Jeogeot is now mapped and good for sailing!
Posted in SL
Tags: Control, Rolling Restarts