I ain’t afraid of No Ghost

I don’t set out to rubbish Linden Lab…really I don’t, but they make it so damn easy to do.

The recent debacle of a code-change applied to the RC sims aimed at improving sim crossing (threaded crossing) has proved to be a monumental failure. It has resulted in a return of that bane of our Second Lives, account “ghosting”, where a sim server never completes the hand off to another sim, and if the account disconnects, as happens in a good proportion of cases, the server does not complete the log out.

This has resulted in hundreds of avatars seemingly online at sim edges, yet with their accounts actually in a “limbo” neither on nor offline. The account cannot be logged in since it is not offline to the login servers, but neither can it be “nudged” offline by a payment or IM sent to the avatar, which was the previous “workaround” if the sim could not be restarted.

Now, the root cause of this was an appalling failure of the QA group at Linden Lab to pick up this aspect of threaded crossing long before it was rolled to the RCs last week. That failure alone is unforgivable, since the change in account handoff MUST have been obvious to anyone trialling the code. So, the flawed code was rolled to the RCs.

That was bad enough. What transpired simply made a bad situation worse. Since the numbers of “ghosted” avatars did not seem to build up until Friday 27th (though some were affected as early as Wednesday 25th), the Lab has taken no action whatsoever to rectify the situation. Indeed if it had not been for the sterling efforts of a number of residents, Marigold Devin at the forefront, in posting to the Server Blog on the 26th that region crossings on LeTigre and Bluesteel RC sims were causing issues, I wonder when the wheel really would have begun to roll. What then appears to have happened was that the majority of the Lab went home on Friday evening and simply forgot about the issue.

At first basic accounts were trying to raise tickets but the majority were either ignored/misunderstood by support or given useless advice about cache-clearing in the “cut and paste” fashion that so much of support appears to rely upon. It was not until late in the UK day on Saturday that any advice that could be deemed as useful appeared on the Grid Status page, and while some avatars have been freed as a result of sim restarts there are still many unable to use their accounts as I type this on Sunday evening (UK time).

This is a comprehensive failure on the part of Linden Lab of just the sort that I have been fearing for some time now…Yes it could have been worse, but what I fear most is that the Lab will not learn from this.

Two things need to happen:
1) QA need to be brought together and collectively kicked up the backside and told in no uncertain terms that if such a thing recurs, a lot of jobs will be lost.
2) Support needs to be re-organised and re-educated. It is simply inadequate for its purpose as it stands and was comprehensively shown up by SL residents.

Two things…two relatively simple but probably costly things. Do I expect them to happen? Frankly no, I don’t. I ain’t afraid of no ghost, but I am afraid that Linden Lab is determined to strangle itself in the medium term. Secondlife is amazing, it is the best virtual world and certainly the biggest by far. How much longer that claim will hold true is the 64000$ question.

Edit Tuesady Feb 1st prior to Roll on LeTigre:
There has been a very great deal of thrashing around over this ghosting issue, and on several occasions I have been told that there are simple and effective remedies for the phenomenon. This has not been borne out in practice for a small but significant number of accounts. The issue is complicated by the fact that in some cases both the region an av was leaving AND the destination sim need to be simultaneously restarted in order to free the “stuck presence”. This is the principal issue surrounding the “Threaded region crossing code” and its failure – avatars never get properly handed off by the region they are leaving. I suspect asset server involvement here as well.
This fact does not seem to have been understood by the people running SL support (nor by some Lindens, which I find worrying). The fact that Support have been asking basic accounts to use a ticket form not available to them is another example of incomplete training of Support staff.
It may well be that today’s restarts on RC sims will clear out those remaining ghosts. The proof of that will be in the cessation of posts to the JIRA SVC-7632/VWR-28209 and the Server Blog.

~ by Ayesha Askham-Ezvalt on January 29, 2012.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: