Ayesha Askham

OK Maestro This week's Restart had Woods of Heaven restarted to sim 9128, and we have had issues again. First symptom I noticed was a higher than usual sim ping of just over 200ms, vs c175ms last week on sim9753. Tonight the ping is over 260ms and we are seeing packet loss of above 1% as opposed to a normal situation on no loss whatsoever. I TP'd to a couple of other Main Server sims, and while the ping was a little elevated it was nothing like as high. Last time I saw this phenomenon was when we were on sim 9078 after a restart last week to 9753 all was well. I wonder if the sims are on the same host or in the same data-centre? 9753 was fine. Yes, ping was a little higher in the evening but nothing major less than 10% - maybe that is at the other data-centre, or at least on a better-performing host machine? I am not planning to add to the already open JIRA on this, but LL need to sort this connectivity issue out. ETA on Wed 6th (GMT): Having restarted the sim to 9133, the stats were back to normal, ie no packet loss and sim ping 175ms+/-10 from over 250ms. Now this sort of difference is critical to those of us outside the USA, and would be bad for anyone with a normal ping of over 100ms. It is clearly internal to Linden Lab and its immediate service providers, so it is up to LL to get this continual failure sorted out. 2nd ETA: OK, so now the system has made a liar of me, sim ping suddenly shot up 50ms in the last 30 mins. What on EARTH is going on? 3rd ETA later the same SL morning: *sigh* SL is making a real mug of me...now everything is back to normal, so I have no clue as to what is causing this.

Mr Nicholai That may now be so. It wasn't always and I had not realised that some sims had changed server. Anyway I referred to Blake "and adjoining regions" and there are certainly several RC channel sims around Blake. This is irrelevant, since the issue of getting one server to hand off smoothly to another is independent of the sim software. The fact remains that crossings that do not involve a teleport are largely borked these days and I see no signs of the matter being addressed and corrected by Linden Lab.

Malate Oh lor'. Yes...I didn't see that paragraph. OK I'm an idiot and possibly also stupid. Sorry! :smileyembarrassed:

On or off...which is it? Nearly 9am SLT and the Scheduled Database maintenance slated for 7am on the GSP on Monday has still not been announced as started. Now either it IS being done and the GSP simply has not been updated, or it has been cancelled/postponed and the GSP has not been updated. Considering that the restart day for the Release Candidate channels was changed to Tuesday to accommodate this, fouling up folks' schedules (OK only us users so who cares) it would have been polite to tell us if this maintenance was or was not happening as scheduled. It is only common courtesy. ETA OK 10am and still nothing. This is really discourteous and darned irritating.

Caleb It hardly seems worth reiterating, though I will, that sim crossings in the Blake Sea and adjoining regions are still awful. Roughly every third or fourth crossing resulted in both myself as passenger and my partner as driver being booted off our slow moving balloon (prim) and deposited at a sim border. I saw several other craft careening pilotless across sims. It seemed unconnected with Channel so I cannot point to a specific RC channel or Main Server as culprit. My guess would be that sims on the same host are passable, sims on different server hosts in the same datacentre are "iffy" and ones that cross to the other datacentre are almost certain dumpsters. It really is high time that this issue was sorted out and eliminated. Sailors and aviators in SL are getting a really raw deal at present.

Jeez Maestro....are You stalking me?? :smileytongue:

Maestro Yep, whoopsie...."They're back!" As to the spammy posts they are now gone too. Thanks....oh...and by the way the Restart warning did get sent today and the new sim seems fine this time. WTG! :smileyhappy:

Maestro I see that the Main Server Restart notice has vanished from the Grid Status Page...Ninja editing or a whoopsie? Also there appear to be numerous forum postings referring to an Escort Agency in the Indian Subcontinent - has the site been hacked??

Kei You don't say whether you are hard-wired or wirelessly connected to SL, or what your set bandwidth is. If you are hard-wired and have either a cable or fibre-optic connection you might try lowering, or even raising, your bandwidth and see if that has any benefit. Since you have already disabled http texture get, I think we can rule out your router/modem, so apart from fiddling with your bandwidth, I cannot suggest anything else. One other thing you might try is to reduce the graphics memory allocation (Prefs>Graphics>settings) to 80% of your current setting and see what that does, though I would have expected a GTX650 Ti to be OK on the default setting, and your driver is up to date. Other than that, sorry, I have nothing to suggest. I would say that the issue is more to do with your router/modem than your PC, but I am by no means an expert.

Well, this is NOT the right place, but... I would first use a viewer that DOES allow you to log in and try to find an in-world support group for the viewer, or failing that Notecard the maker of the Viewer, or one of them. This is the Server Forum and not the Viewer forum anyway, and you may get some peer-to-peer help but most likely a moderator will redirect this post to a general discussion forum. You might also find some advice at SL Universe.

Maestro Speaking of things fixed by restarts: The issues with poor TD and script-run percentage on Woods of Heaven as sim 9087 were cured by a restart to sim no.9825, so there is definitely some issue with the sim-server at sim 9087. Or, it might just have been that the other sims on that server-host were heavily laden and taking up too many server cycles.

A bit of forum weirdness, Caleb I saw your post, but it vanished almost before my eyes! Just one issue, Woods is now sim 9087, and while it may not be bad by Mainland standards, its performance is woeful by normal standards, and we are not about to sit by and let this go unchallenged. Just because Mainland sims run poorly does not mean that we accept such instability as a norm for an island region, Homestead or not. I have been in SL 5 and a half years and on this Homestead for 3. We want it running properly, despite Support's comment. Unreliable script-run and TD oscillations are simply not acceptable on a sim that usually runs flawlessly.

Weird Just saw a fleeting post by Caleb...Woods of Heaven was sim 9786. Just to be clear I got the standard log-out message as usual so it was NOT a viewer crash.

OK after restart: Sim no. 9087, Woods of Heaven, TD and script run are unstable, TD 0.90-1.0, Script run 95-99.5% Sim FPS and physics FPS also sporadically dropping to <42fps Sim No 9742 Lost Island, TD and script run are stable: TD 0.999-1.0, Script run 99.8% Both on new Server Version. This suggests a problem with sim 9087 since Woods is usually at least as good as Lost Island, both are Homesteads and lightly loaded script-wise. Performance of Woods is inadequate as things stand. Please correct this.

Maestro This is a simple, heartfelt plea. PLEASE can you sort out the Server Code so that avs get a warning before being summarily booted out at region restarts. Yet again I received no countdown warnings at ALL before being disconnected just now. I thought you had this issue fixed a few versions back, but it would seem that I am mistaken. I follow the advice code so nothing will be lost it is just damned inconvenient and above all just plain rude. :smileymad:

dd I am not one to rant and whinge, but some of Lydia's points are quite valid. There is without doubt some issue regarding server function at Linden Lab. The variation in sim ping and resultant packet loss when pings exceed 250ms is something which I have experienced first hand. This phenomenon does not seem related to particular channels, since Main Server sims can vary by up to 100ms, for no apparent reason. Certain sim crossings when undertaken on vehicles are quite good, even between different channels, but some, and these can be between two Main Server sims can be dreadful and may result in either mangled data crashes, unseating of passengers, camera angle distortion or other occurrences...I have seen all of them at one time or another. That there is also a server latency issue with Abovenet is certain (I have data), but whether there is also an issue with either LL's internal comms or the first stage of data transmission from LL I do not know nor can I determine. I have raised an LL Bug report BUG-4025, and I know there are others, but of late I have seen little or no action on this. The US admin shutdown has complicated tracing issues, but I believe they still remain unsolved and unsorted. There is no doubt that on some sims as soon as one lands after a TP and for up to 5 minutes after rezzing, there are are waves of data being sent to the client (seen as periodic rises in bandwidth being used) irrespective of draw distance. That is not normal but what it signifies I do not know. Lydia There is a fundamental disconnect between us the users of Secondlife as an environment and the developers at Linden Lab. They simply do not wish to be bothered with ironing out what they see as minor failures in their work. Certain Linden employees and ex-employees who have tried to get them to focus on getting SL running properly liken the task to herding cats or sieving a gas. It is almost impossible, despite its significantly negative impact on both us, the users of SL and its perceived image in the wider business.

Curiouser and curiouser..... Sometime in the last 24-48 Hours Virgin Media rerouted my connection to SL via servers in the Netherlands...a company with initials ams.nl (?). Currently I see a slightly raised but consistent ping to SL, though sim-to-sim variations are still evident.

Dear All Being in the UK with a moderately scripted (70) avatar with a 45K inventory, it would be expected that I might have issues. Other than the Abovenet server latency issue that magically disappeared when the US admin shutdown began, or the month of October started (you choose), I experience no issues this week that were not present last week or before. Whatever the cause of the outrageously high pings to Europe, the UK and Australia (those are the regions that I KNOW had issues, there were almost certainly others) I'd say SL was running reasonably well, for a given value of well. The http issues, sim crossing issues, interest list issues and materials issues are all known, and will be fixed no doubt in the fullness of time. The issues that those such as Lydia suffer probably will not until Linden Lab also employ psychiatrists.

Norton I would not ordinarily disagree with you on most topics in SL, but in the last few months, with every rolling restart on Main Server I have found our region to be on a new server. Not occasionally but EVERY week, a new server. This situation was not always thus but it is so now. Yoriko Is your sim ping still high? As of yesterday at about 9am Pacific mine returned to normal values everywhere I went. I do not know what changed...maybe the US Admin shutdown is responsible for lower net traffic? :smileysurprised:

Yuriko You are almost certainly right, it is a problem between the first of Abovenet's datacentres where the cable resurfaces in the USA and Linden Lab. Detailed examination of tracert suggests that the culprit is Abovenet's datacentre phx2. Now Linden Lab also have a datacentre there...so...? Following the creation of https://jira.secondlife.com/browse/BUG-4025 and some hunting about for Maestro Linden, I think it is possible that The Lab might weigh in with Abovenet's parent ZayoNCC, and possibly at least find out why this situation has arisen. Whether they will then solve it is another matter altogether.

Theresa You are essentially correct in that VM does indeed use Abovenet as its peer partner, as do many European ISPs. Hence many of us are seeing the same issue. Whether changes in the volume of data streamed by SL servers is implicated is up for debate and is certainly not proven at this time. VM have a trouble ticket raised now about this. Thanks to Izzy Linden and also Monty there is now an SL Bug report on the matter. https://jira.secondlife.com/browse/BUG-4025 Where we go next with this issue is rather up to Abovenet's parent, ZayoNCC.

I normally do not post comments like this but tonight took the biscuit. I posted accurate technical details of the sim ping issue that users in Western Europe and the UK are having with abovenet links to SL as a ticket to Support. The advice I received? To clear my cache and reinstall my viewer. !! From a Linden!!

Latest info Have made contact with Zayo, the Abovenet parent company, but they are asking for account information, which I have no idea how to provide. I assume my ISP would have such an account, but so far I have not been able to obtain it. I do not know how to progress this matter further. ETA Friday 27th: I have raised a ticket with Virgin Media today quoting some info given by support at Zayo (Abovenet), so we will see what happens. I know that not all affected users are on VM connection, but I bet most, if not all, use Abovenet to reach SL. I do wonder if the recent increase that we all probably see in sim-to-client data streaming is causing one of the servers to throttle the connection? That implicates recent SL server-software "improvements". @Yuriko: I do notice that once your connection reaches abovenet at uk.above.net, the pathway is identical to mine.

Adding to Wolf's comment: Last night (UK time) on Blake was just about the worst sim-crossing experience I have EVER had in SL. Like Wolf I must have failed to recognise the problem when this software was on Magnum. I was stuck ghosted on nearly every sim crossing on Blake, camera view was corrupted several times and ultimately my viewer crashed. Upon attempting a relog to the same point I got the message "Login Failed. Agent not on sim. Please try again in a few minutes". I eventually managed to log in at home, but the whole process was appalling. Coupled with the currently atrocious performance of the transatlantic/USA servers of abovenet and some undefined networking issues between LL datacentres, in the USA business day, anything that involves more than sitting around chatting is almost impossible for users in Western Europe. Linden Lab this must stop now.

Please see the ETA 4 on my previous post. Perhaps someone could explain how this could occur?