Slow Texture loading in recent viewers (1.21 and later)

Okay....now I feel really blonde and useless. I guess I did have a false impression of helping in some small way. Thanks Sindy for letting me know I have been wasting my time ...I will not make that mistake again I have come to the conclusion that there are just too many personal (resentments/problems/feuds/dislikes) on this forum for it to actually be good for anything but fueling them. Too bad

- what country are you in and who is your ISP?
Germany using T-Online

- do you use a wireless network at home?
No

- shut down your computer and power off (actually unplug the power cord) ALL your networking stuff for a minute. Power back up and log in.. Any better?
No change

- do "ping -t a.b.c.d" in a command window, let that run for 5 minutes or so, then press CTRL-C and post the last 5 lines of the output here.
--- 216.82.19.229 ping statistics ---
300 packets transmitted, 300 received, 0% packet loss, time 300314ms
rtt min/avg/max/mdev = 137.622/140.754/168.235/3.838 ms

- do "tracert a.b.c.d" and post the result here. Edit out the first 3-4 lines, which may contain personal info
4 f-ea5.F.DE.net.DTAG.DE (62.154.16.165) 20.782 ms 21.885 ms 23.293 ms
5 te-4-4.car2.Frankfurt1.Level3.net (4.68.110.253) 25.601 ms 26.028 ms 27.623 ms
6 vlan99.csw4.Frankfurt1.Level3.net (4.68.23.254) 42.696 ms 29.860 ms 30.201 ms
7 ae-92-92.ebr2.Frankfurt1.Level3.net (4.69.140.29) 20.048 ms 21.207 ms 21.391 ms
8 ae-10-10.ebr2.Washington1.Level3.net (4.69.141.110) 109.882 ms 106.778 ms 106.745 ms
9 ae-92-92.csw4.Washington1.Level3.net (4.69.134.15 107.997 ms 106.092 ms 106.996 ms
10 ae-91-91.ebr1.Washington1.Level3.net (4.69.134.141) 115.179 ms 109.867 ms 108.454 ms
11 ae-2.ebr3.Atlanta2.Level3.net (4.69.132.85) 119.588 ms 131.168 ms 131.320 ms
12 ae-7.ebr3.Dallas1.Level3.net (4.69.134.21) 151.553 ms 141.514 ms 140.863 ms
13 ae-63-63.csw1.Dallas1.Level3.net (4.69.136.154) 156.633 ms 154.339 ms 154.261 ms
14 ae-12-69.car2.Dallas1.Level3.net (4.68.19.4) 141.948 ms 142.025 ms 143.056 ms
15 DATABANK-HO.car2.Dallas1.Level3.net (4.71.170.2) 144.287 ms 141.049 ms 141.972 ms
16 ae_cw_10g.databank.com (63.164.96.62) 142.271 ms 140.603 ms 141.621 ms
17 sim2986.agni.lindenlab.com (216.82.19.229) 141.642 ms 138.164 ms 143.041 ms

The point was that if you're going to do a survey, do something useful.

A survey on what speeds your ISP says you're getting is like a survey asking what color your computer case is.. There are better questions to ask if you suspect network problems - keep reading for what I would have asked.

That's not 5 minutes worth!!

A 10-15 second sample only shows you a very brief snapshot of how your connection to the sim is doing.. Really take 5 minutes and just let it run in the background while you're in SL. Or even 30 minutes or an hour if you want.. The goal is to see how it does over time.

edit: ok.. ty..

I did not "hide" an "attack" at all Sindy. After this user repeatedly ignored all other comments, pleas and expressions of anger and continued to enforce his whims on the JIRA (resulting in a Linden stepping in and deleting massive amounts of posts) I laid the FACTS out here for everyone to see... that the user abused other users, angered them enough to provoke responses that got them banned... has done so BEFORE and not learned his lesson from that, and basically derailed a valid thread until the Lindens had to step in and majorly moderate it. I "hid" nothing. I posted facts and pointed the blame where it belonged. That's called "standing up for what's right". Sorry that offends you. Not my problem.

It's pretty obvious to me that the problem here is one of you having issues against me personally (for whatever reason... who's actually hiding true motives?) So you'll pardon me if I ignore future such posts by you and suggest that you focus on contributing to the thread and the issue at hand instead of trolling, because frankly madam, your rants here are far more extensive, numerous and objectionable... and it is you, not me that is derailing even this thread. So please, take the attitude elswhere.

Sindy... this information is useful because it provides statistical data on individual user experience. Is the user using DSL or Cable? Is it a fast connection or a slow one? Is there significant statistical correlation in the differences between DSL and cable connection? While I agree that connection speeds are LIKELY irrelevant, ignoring data because of pre-conceived expectations is the best way I know of to miss out on unexpected results.

The simplicity of that survey takes into account that the majority of SL users are not techs, will not mess with lengthy surveys, and may not even be able to answer some of the questions already there.

It is VERY apparent from posts thus far that "user interaction" for texture loading works with some people-- and with others it doesn't. It's not black and white, cut and dried "user interaction helps or it doesn't". The question that thus remains is WHY does it help some users while NOT helping others. That's the purpose of collecting raw statistical data... to try and determine that reason.

Wayfinder: Your two last posts are irrelevant to our problem as well. If don't stop this childish argue, all we can achieve is that LL is still ignoring us.

And I have to say: I cannot tell why they should not ignore us, such childs as we proof to be here!

If you want to be productive, be it, well appreciated (as you can see with my participation). If you can't be productive, then keep your fingers still, because everything you can post them will be contra-productive.

I PRAY that this now is the last post about this, because if I read anything on this level here again, you can not count on my help and participation anymore, as well as surely many others will not help with your surveys when they see it's just a flamewar thread!

/me makes the universal "blah, blah, blah" motion with her hand and adds to her ignore list..

Daniel, I agree with your thoughts and desire to make peace. That isn't accomplished by further ranting and insulting others, is it? But while you're pointing the finger at others, I might point out (and forgive me) that you've had your little jaunt sparring back and forth with Sindy. You said your bit and now you're trying to silence others? Seriously...

A note on my initial "rant". For the record... this forum is NOT for the specific purpose of discussing the issue at hand. If people want to discuss pure specs and tech... that's what the JIRA is supposedly for, is it not? This forum was pushed here by a Linden due to an abusive user derailing the JIRA thread. To expect that users are not going to comment on that experience and further-- to try to moderate this forum to strictly techinical issues... is overstepping.

I defend your right to speak your opinion, but not to cast blame at others while posting in the same manner. Your chides and calling others "childish" are not exactly "tech" are they?

I want to say something here about my initial "rant". If you haven't noticed, I'm the one who originally posted the JIRA that was derailed by both Linden Lab and Gordon. I filed that JIRA after MONTHS and YEARS of trying to bring this problem to the attention of Linden Lab by other means. Linden Lab derailed that JIRA by trying to silence me from debating with a tech who was providing questionable responses (and it amazes me how many people swallow his statements hook line and sinker when to both me and Balpien Hammerer the loopholes are obvious. Go figure).

They further derailed it by making the same claims and demands over MONTHS of time (these are separate issues, feed us more data, we don't have enough to work with) while hiding the fact that they've been aware of these issues for YEARS, were told of these issues long before the JIRA even existed, and those issues have gone largely ignored until I did post that JIRA. My "rant" exposes all that... and if anyone doesn't see the validity of doing so, then they are welcome to return to burying their heads in the sand. Standing up and saying "Linden Lab is not being honest with us" is not a "rant"... especially when accompanied by factual information.

Then, when Gordon successfully derailed yet ANOTHER thread (is the guy a secret LL employee?? One has to wonder LOL) ... how does Linden Lab react? Instead of removing the flamer posts that derailed the thread, they find it appropriate to go back through the entire thread and remove ANYTHING that can be considered even moderately opposed to Linden Lab philosophies... going so far as to edit even my original post and my debates with Dan Linden over technical and Linden handling aspects of the issue. Now many people, Daniel, would lablel such as "censorship" or "JIRA Nazi" activity-- massive deleting and editing of posts, regardless of validity to the issue at hand.

Linden Lab consistently insists on "just the facts" in the JIRA system. Those "facts" fail to take into consideration the "human factor" of their customers... the degree of damage those bugs are causing, their effect on the social and financial status of the grid, or the fact that these bugs have been reported for YEARS without response. Not everything reported in JIRA can be parred down to mathematical equasions. To prevent users from stating their opinions and feelings in regard to those bugs is, as I've stated, insulting to the customers.

CUSTOMERS IN GENERAL ARE NOT TECHS. They should not be expected to file JIRA reports in some straight-line goose-step formation. They are people with feelings and wants and needs and sometimes when a bug affects them majorly, they are going to be irate. Linden Lab is expected to have thick enough skins to deal with such without flying off the handle and banning people without prior warning or appeals.

IMO the company policies evidenced by the actions on that particular JIRA post are inexcusable customer support. They try to quantify human experience and emotion, they try to force customer experience into a convenient and comfortable Linden Lab cubicle mold. I think it a disservice Daniel, for a user to try and enforce such policies here.

Suggestion: realize that these are FORUMS and not the JIRA and realize that people do have the right to state their opinions, expose truths, and even rant if they are so inclined. This is a discussion and debate area; it is most certainly not the sterile white-scrubbed and (in the opinion of many) extremely unfriendly and unproductive environment of the JIRA

If these issues cause you to decide to not contribute in this forum Daniel, that is your privilege and decision... just so long as you don't attempt to tell others what we can and can't post here within the guidelines of the TOS.

In these forums I exposed abuse both by Gordon and Linden Lab that caused these issues to be moved to these forums in the first place. Since I have not commented on that JIRA for months, none of that was MY doing, was it? And even now I've been trying to collect some small amount of data in an effort to determine if there is an unobserved pattern in texture loading issues. But that's likely not going to work because Sindy found it proper to try to derail even that rather than offering a constructive alternative. So frankly Daniel, if there's anyone here who's butting his head against a brick wall of uncooperative people... I'd say it's me.

People can participate or refuse to participate in that survey as they choose. I've dealth with Linden Lab for years over this very issue... and you see how they've responded. The problems still exist and are even growing worse. You think I'm not handling this correctly... now that the tough pioneering and road paving and foundation work is done, YOU deal with the issue for a while. Washing my hands of it. My group has other plans altogether and I have no intention on helping with this issue any further.

[.......... deleted TONS of text ]

I brought Sindy to sattle down without pointing at you and I thought you could now silence this topic without loosing your face. Seems you are not adult enough to do that. Sindy is, as she proved by nearly completely ignoring the unproductive posts of you.

Really, I don't have time to read this novel above. I will search for another thread where hopefully all participants can focus on the problem instead of their ego.
I can cope with the texture loading bug, because selecting objects works for me. You don't want my help, you don't get it. I will think about unwatching and unvoting the jiras. Have fun with your grey textures. kthxbai

Daniel, that you don't see the irony in this post is frankly amusing-- and amazing.

So you don't bother to actually READ that "tons of text"-- you just decided to cop an attitude at the first paragraph and ignore the rest? If you "don't have time" to read the forum posts you're responding to, most would consider it wise to withhold a response.

Please... DO go to another thread. That you would have the gall to call others childish and egotistical while posting rants such as exhibited above is just too much. Seriously, you ceased dissussing the "facts" about a page back and every post by you since then has been chiding and insulting others-- while telling them in the same breath they should't be doing the same. I've insulted no one. I have exposed factual activities. Remove the rafter from your own eye, sir.

I've dealt with troll posts and troll "mobs" on these forums for longer than you've been a member of SL. Reading back in the archives of those posts proves most enlightening and quickly exposes ulterior activities. When (if?) Linden Lab ever does get this problem fixed, I'm fairly confident the facts will prove equally as enlightening here. No brag, just fact-- past forum threads have wound up with my postings proving spot-on accurate-- and the trolls a bunch of blustering knobs presenting nothing to the discussion but pure attitude.

I don't care what you personally think about my presentations here... I'll challenge you to disprove even one statement made with the same facts and evidence in which such are presented-- and without resorting to labeling and insulting to accomplish your goal. But to do so... you may find that first you actually have to READ the posts instead of just blindly responding to imaginings of what you think they say. There is pertinent and valid information there-- if you'd just "take the time" to read it. If you choose not to do so... please spare us further attitude.

Don't feel bad, Daniel.. You're not alone in wanting things to be productive or in being annoyed at people who have other agendas. You're also far more polite than I.

Your network stuff looks pretty damn good for being 1/3 way around the world, BTW. I don't think it's safe to declare "probably not a network problem" with only one person responding with actual data (vs quoting marketing data from their ISP) but unless several other people show tracerts with the same hops close to their homes, network probably isn't the source of the problem...

To bring this forum down to earth... and to demonstrate what has been discussed thus far in this thread... this forum is entitled

"Slow Texture loading in recent viewers (1.21 and later)

Who came up with the idea that this affects only viewers 1.21 and later? It wasn't me... and I posted the original JIRA.

The title and concept itself is misleading and is NOT the title I originally placed on that JIRA. Who has the right to change and edit what the original poster put on a JIRA? That is misrepresentative.

As evidenced both in my original post and in the discussion that follows, I brought to light that this texture issue has existed for quite some time but has increased recently. A quote from that original post is:

"Similar reports have been issued before, but 1.20 brings this problem to an all-new level.."

So how did this go from such a report to dealing specifically to "viewers 1.21 and later"?

This question is especially posed since it was later discovered and proved on that JIRA that the recent slowdowns affect PRIOR viewers as well... .that the slowdown is not caused (primarily) by viewers but rather by obvious re-loading of textures instead of properly caching them. It's not the viewers specifically (although those may well be involved) but rather the way texture transfer is being handled across the board... affecting ALL viewer versions.

This is what I talk about when I discuss improper censoring and modifications of JIRA reports. I don't care what people wish to add in comments after the original report... IMO allowing others to change the content and even the TITLE of an original JIRA report is improper. Only the original poster should have the right to edit the title and/or original post. Anything otherwise is to post something in that user's name-- which is fraudulent. That's not what the user originally posted.

Ah I found the ignore list to be great to make these huge novels to hide. Perhaps this might be a way to still continue being constructive here.



Hmm I could imagine that could be the cause of the difference in our experience. Because as far as I remember, the select/hover/right click trick does not work for you, whist it works for me, isn't it?

Could you post your network details, perhaps we can find some significant difference?

Yes. Ignore is a great forums feature!

Nope - I'm not really having problems at all.. Textures aren't uberfast but they're not bad and will download most everything in a busy area within a few minutes. Hovering will speed up individual things.

I'm in the eastern US and don't use wireless on this PC.

My tracert to the sim you were using:
4 21 ms 20 ms 20 ms 166.90.136.33
5 24 ms 20 ms 33 ms vlan52.csw2.NewYork1.Level3.net [4.68.97.62]
6 24 ms 20 ms 33 ms ae-73-73.ebr3.NewYork1.Level3.net [4.69.134.101]
7 62 ms 56 ms 68 ms ae-6.ebr2.Dallas1.Level3.net [4.69.137.121]
8 57 ms 66 ms 55 ms ae-92-92.csw4.Dallas1.Level3.net [4.69.136.150]
9 57 ms 56 ms 56 ms ae-42-99.car2.Dallas1.Level3.net [4.68.19.196]
10 57 ms 56 ms 56 ms DATABANK-HO.car2.Dallas1.Level3.net [4.71.170.2]
11 56 ms 57 ms 56 ms ae_cw_10g.databank.com [63.164.96.62]
12 56 ms 56 ms 56 ms sim2986.agni.lindenlab.com [216.82.19.229]

ping -t for not-quite-5-minutes

Packets: Sent = 219, Received = 219, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 54ms, Maximum = 59ms, Average = 55ms


The thing to look for now, assuming network is still suspect and that people are having the same problem, is things that are similar. Like: "all the people having problems are in Germany" or "all the people having problems are in Europe" or that they all come through a Washington DC router, like you do..

Ah okay, remembered that wrong or got this puzzled. So the first hypothesis out of that info would be: Good Internet connection waters down the symptoms. You have way better connection than me and have lesser problems.

Perhaps one of the others could state their rough position by country and give the result of a some-minutes ping and the tracert to 216.82.19.229?

Especially now users with heavier problems would be interesting, those who do not benefit from selecting/clicking objects?

That would point to the two problems
1.: Wrong texture priority (that's what I suffer from)
2.: (Assumed) lower network quality makes texture loads stall forever.

Belle, perhaps you suffer from second problem? Lower network quality doesn't mean yours could be real bad, it rather means, from time to time perhaps a packet drops. Normally a correct network software resends this packet, you don't notice that in browser or such, but there could be a bug in SL with resending or such.

I'd actually suggest they do the Help->About SL and get the IP of a sim their having problems in. LL has locations in Texas, California and Arizona. If people report problems in each location, that'll be another data point..

Admitted, this makes sense.

This has always happened since I started SL. It doesn't happen for every texture, but does happen very often, particularly but not limited to textures on avatars.

When I see the textures loading they start off blurry and gradually sharpen up, but then a lot of the time the texture will go blurry again and then slowly sharpen up again.

Even if I allow a region to fully load, then log off, then log back on, some textures will load and then go blurry as if they are downloading again.

Could anyone explain what is happening? Is the viewer downloading hundreds of textures multiple times even when I already have the texture? It would explain why everything downloads so slowly.

There are issues with the 'discard levels' that I know LL is working on now.. http://jira.secondlife.com/browse/SVC-3146 is one and Simon mentioned another related one last night at Andrews office hour. Simon seemed pretty excited and I'm looking for textures to be a lot faster before long - it'll be a new viewer update, though.

I think the problem your seeing is different than the one talked about in this thread. The problem here seems to be that they're getting nothing, not even partial fuzzy textures...

LOL I alwasy have to chuckle when people not only figure out how to use the ignore button... but find themselves compelled to publicly announce the fact they've done so. LOL

Such activities only limits THEM reading valid response to their claims-- but by no means blocks others from reading valid responses that factually disprove their claims. Thus, they fail to learn from others and continue on the same incorrect logic path. The rest of us learn while such ones remain unaware the points being made have already been invalidated. That's just poor.

Regarding the somewhat extensive network/ping posts... they aren't really applicable to the issue, are they? It really doesn't assist the users (us) nor Linden Lab (them).... because what is being focused on is exclusively ping rates and network performance... which has little or nothing to do with proper transfer protocol and handling of cache issues. What little it does have to do with-- and the extremely few number of people such might affect-- the issue is being overly emphasized, as such is pretty obviously not the primary problem here.

If ping rates and network performance were involved... the entire Second Life experience would be a shambles (well, in truth, it is, but not in the same manner that texture loading is a shambles). Failed group chat (database and internal network issues), lost inventory (failed database), falling through floors (recent and failed coding) etc really don't have anything to do directly with ping rates... nor does texture transfer.

Look, if ping and network rates are the issue... then simply switching to a non-recurring protocol (such as http instead of UPD) would totally fix the problem, wouldn't it? Ping and Network issues in truth have nothing to do with the same textures reloading over and over again instead of pulling from cache.

So really, this mutual admiration society is not actually addressing the issue at hand... namely, INTERNAL Linden Lab coding, system structure and protocols that simply are not doing the job. It is yet once again focusing on the CUSTOMER instead of the SOURCE of the problem.

The only reason the prior survey was presented was to factually establish whether or not the issue is CUSTOMER side or SERVER side. I fully expect the results of such will prove to be server issues. But that's a moot point since apparently not enough people use the forums or wish to get involved to establish a valid population database. So at this point... further discussion here would likely be a mere repeat of what has already been fully discussed on the JIRA. If those with "textures aren't uber fast but aren't bad at all" experiences had chosen to actually participate in that survey rather than copping an attitude, we might have actually had some valid data. Attitude in place of cooperation tends to produce zero results.

A lot of the really VALID points and posts in the orignal JIRA were removed and placed here in random order. That pretty much negates and invalidates the JIRA process... and puts all the work customers put into providing feedback there into the "royal waste of time" bin.

Hmmmyes, but perhaps the issue described could be this cache aggressivity problem, what do you think?

The problem as far as I understood, is that the viewer discards textures from the local viewer cache too fast. So perhaps here we have a slow/stalled texture loading because of network problems, and before the texture could be fully loaded, the cache entry expired and the viewer starts downloading the texture again freshly.

The viewer cache algorithm needs to be re-evaluated.

A squid-style pure file cache for textures would avoid a lot of the problems that the current virtual file system font-ended cache suffers from.

Barking up the wrong tree.

While that may be true in SOME instances... for access to be that slow...

* Second Life would be continually and totally stalled for that user-- virtually unusuable

* Inventory would not be accessible

* The majority of chat would totally fail

* The problem would be exclusive to those limited numbers of users instead of affecting HUNDREDS OF THOUSANDS of people on the grid.

Look at the number of votes on this issue. Look at the number of people watching this issue. Then apply common sense. It's NOT networking issues. It's how Second Life is coded and internally structured / organized to react to standard user access, as well as some rather serious protocol and design flaws. Seriously, think this stuff through folks.

I know of course the poster is not going to be able to read this factual feedback because she and her "partner" decided to ignore the original poster of the JIRA, but that's their problem. If they want to run away from the facts, let them continue barking up the wrong tree. The rest of us are informed. It's NOT personal user network issues. While the UDP protocol may indeed be playing havok because of ping/dropped packet problems... that fact has LONG ago been recognized. Putting a non-redundant protocol in place will likely erase that issue completely. What IS the problem is far beyond that... namely, cache handling, a major delay-of-service attack, and generally very poor texture transfer design... as well as 5 years of failing to fix a known problem.

I agree Argent, that does need to be looked at as part of the problem. It's apparent than in the last 2 or 3 releases Linden Lab majorly changed the way the system downloads and caches textures. Definitely coding issues, probably in both the client and server-side software. In fact, in the JIRA several users not only recommended totally revisiting the cache algorithms... but expanding available hard drive cache as well. When folks have half a terrabyte or more of HD space on their systems... we can afford the room.