Question for ya JP... What is the current value we're using here for the board "time-out for cookie" setting? And, has this changed from what it (or the equivalent) was under teh old board software?
Printable View
Question for ya JP... What is the current value we're using here for the board "time-out for cookie" setting? And, has this changed from what it (or the equivalent) was under teh old board software?
There isn't any time out, as far as I know. The board never logs me out until I log out manually. It'll keep me logged in for weeks. You're getting sessions interrupted?
I think I misunderstood you. Are you referring to the time before posts/threads are marked read?
My quesion relates to just an idle curiosity of mine, not a problem.
Since upgading to the new board (starting pretty much immediately) there has clearly been a rise in the number of online users (members + guests), and at most times of the day. That there would be a rise doesn't surprise me necessarily. But the magnitude of the rise just seems way to high (its like doubled or better). As a result I have been wondering if there was something that changed with the advent of the new board vs the old. I'm not implying a bug, just either some setting value change or change in the way the statistics/data are gathered/measured under the new board. To this end, it is/was my understanding (and it may be wrong) that one of the board options that could effect (among other things) the count of online users was one called "Time-Out for Cookie", that sufficiently high values for this setting could inflate the statistics. So, I was just trying to find out what our current value was, as well as what it (or a like setting) was before the board change.
Yeah, I've noticed that too, we all have. I assume bots account for some portion, since this version is far more spider friendly. But I've just reduced the timeout by a huge margin, so that should speed threads being marked read, as well as the lingering of who's online, when they no longer are. I don't know what the old board setting was.
How about if I just close my browser and shutdown my PC ?
As far as VDr is concerned I'm still logged in.
How does it know to kill my session ?
Every forum (and other sites as well) use what's called a session cookie. The "session object" is defined as how long the browser is open to a page. The server reads the cookie at set intervals. When it no longer can, it knows you've left the page, hence, session ended. The setting Doc refers to determines how long after that your status on the board would change.
Exactly JP (the session cookie stuff).
It is indeed odd. Like I say, I don't think it is any sort of bug, just something different. Ya just can't help but conclude though that the rise in the numbers was/is just too high, too fast. And based on this it seems logical that it comes down to something being counted/measured (generically speaking) differently than before.
Prior to the upgrade in some of the discussions in that feedback thread I started we talked about us vs other boards in terms of online users. One of the things that to me stuck out at the time (though wasn't really delved in to) was that often the number of "guests" showing up was quite high on these boards, often VERY high. At the time these boards were running the newer software relative to us. Now we are running a comparable version of vbulletin. And, now all of a sudden our online (particularly ""guest") numbers have jumped. Certainly makes one wonder if there is a connection.
Personally, I just think something is resulting (again, not meaning a bug) in connections to be hung on to or perceived as open longer now, and thus getting reflected back as higher online activity.
Well how long for what I'm talking about ?
Edit: Just read DrMDJ's post and am now :confused: - are we talking about the same thing ?
Well I've just done some investigating and it appears the default setting for cookei timeout in the old version is the same.
Actually Nix. What we really need to know to be able to compare apples to apples in looking at things is the relative "now vs before" information.
How long for the change in status to show? 900 seconds, currently. That's the default for the old and new versions, so I decided to leave it.
Well, if the cookie setting is the same then obviously that ain't it. But it just seems something has to account for this (and it would be an unrealistic stretch to attribute this to any of the search engine stuff). It would be interesting to know.
Well I checked some of our boards that haven't been upgraded yet, and they show 900 seconds too, but I can't state with certainty that this board was previously set to the default, in the older version. I never had occasion to check.
Well in the VDr is a little sick thread on around 22nd March I posted that there were 74,275 members and there are now 74,739 in fact I'm sure there's been an increase of about 60 or so in the last day.
That seems like alot
Well those readings may not be connected to this at all. I know the thread count is and has been well off. There are a lot more posts than that figure currently shows. I don't think it's governed by this setting.
And, it may not even be that value that is behind this. Like I say though, something seems odd. Case in point... A little over an hour ago our online statics were way down, like under maybe 35 including guest, with guests being like 13 or something. Then gradually after a bit the numbers just grew (quickly) substantially, especially the "guest" numbers (it's always the guest numbers that are high). Very suspicious.
Quick question JP.... Were you doing something in the last hour (roughly)? There was a brief period of time where I was trying to reply to this thread and twice after just (literally) a couple minutes I timed out (as seen when I did a preview of my reply) and had to log back in (not happening anymore). It also seemed like out online statics dropped SHARPLY at about that same time period.
41 new members joined on for the period 04/07/2005 until now Nix.
Nothing that should affect that. I changed the display order of the forums this morning, and I just changed this setting we're discussing twice (back to the default now). Don't know if they would cause what you saw though.
If you have multiple threads open in Tabs in FF does that count as one session or multiple ?
Just was curious JP. It was like something caused a "reset" (if ya will) to take place.
Nix, on that "new members" like of things... An interesting set of numbers if ya like numbers (like all statistics, take it for what it is worth): Members joining between 3/16/2005 and 4/8/2005 (I used 3/16 since that's when the board was upgraded as I recall) = 343. Members joining for same period in 2004 = 399.
Anyway... Like I say, the online numbers do seem to give rise to question. Primarily those for "guests".
That would be one session, I believe, Nix. I don't believe it sets a cookie per page, but per site.
I notice that usually at around 4.00pm in australia the number of members might be 4-5 vs guests 25-30 or about 1:6
So when the us and uk wake up a bit more and i should be in bed (it's 12.45am) the current numbers are 22 members 169 guests which is 1:7.68
So.....
Or, are the statistics in any way based on IP (port) connections?
Well I just logged out and opened about 30 threads in tabs and it didn't jump the numbers up by 30.
I tell ya, there's something with that "guest" number. It just seems to run wild. And now, after our experience I go look at a place like techsupportguy forums now and see 114 members and 1380 guests and say hmmm.
Like I said, you would still represent 1 user, no matter how many threads you open. It can identify IPs as well as use session cookies.
This is interesting. The figures seem to be skewing somehow. In checking a couple of other sites I see the same thing. That indicates that it is not the sites per se, but it could be the software the sites are using. Or is it a crawler/search engine?
There is one thing for sure there is an accumulator somewhere that might ought to put in for overtime. :D That accumulator has to release what it is holding too at some point.
How does this impact an advertising dollar that is spent I wonder.
Page loads are monitored elsewhere, I'm pretty sure. I don't believe the forum software is responsible for that.
What we have seen here since getting the software upgrade is entirely consistent with VDr now being visible in the search engines. Because people can now find us, so they come and visit, and some of those guests then join the forum. In fact there is a nice consistency with the numbers, the increase in new members joining is nicely proportional to the guests online. The figures are also consistent with other forums running VB3, and what we have seen happen here is identical to what happened after VBForums upgraded.
On several occasions recently when searching for an answer I've seen Virtual Dr listed in the results, which is something that I haven't seen for a long time. And that is esxactly what we expected to happen after the upgrade, the serach engine friendliness of the new version make a tremendous difference.
As I said, I think there is a suspicious proportion in the "guests" online number. Looking at other forums prior to our upgrade their guest numbers were large, but I did not pay much attention to it. Now that our number has jump so dramatically (and pretty much right from the start after the upgrade) it has to make one wonder. Certainly the improved search engine stuff would be expected to have an effect on people finding this (or any) site and coming here. But I think there is reason for healthy suspicion that the jump is a bit extreme. It is just too odd that suddenly, just about any hour of the day or night, our numbers are so much higher. It can't be that that many people out there are running around using search engines looking for online help forums and then going to these and just hanging out all the time. When numbers look too good to seem true, they often are.Quote:
The figures are also consistent with other forums running VB3, and what we have seen happen here is identical to what happened after VBForums upgraded.
Which other forums, JupiterMedia ones?Quote:
The figures are also consistent with other forums running VB3, and what we have seen happen here is identical to what happened after VBForums upgraded.
As for new members joining, I would expect an improvement post upgrade. However, as I said earlier. If you look at the the period from when we upgraded to today there have actually been less new members join than in the same period last year. Not saying definatively what that means. But it does show that statitics can show a lot of things.
Page loads and that are in the PHP side, if that is what V bulliten is running on. Ip would be listed with the uploads
The jump in guests online did not start immediately, it has been steadily climbing since the upgrade. The average number of guests online were 32 for the 23 days prior to the upgrade, and 92 for the 23 days since the upgrade (all figures taken at approximately 1900 UK time). If anything I had hoped for a bit more of a dramatic increase, given that we are far more visible in the search engines now.
The other forums that I collects stats from include some JM sites and some non JM sites.
In this period last year we had at least as many guests online as we have seen in the last few days, the numbers dropped off sometime late last summer. So I would expect there to have been more members joining this time last year, since we have only got back up to last years guest figures in the last 10 days.
That's a 3 to 1 increase. Just a bit too suspicious.Quote:
The average number of guests online were 32 for the 23 days prior to the upgrade, and 92 for the 23 days since the upgrade
And when you say "suspicious" what do you mean? What are you insinuating?
Just what I have from the start. Namely that there is something in the new board software (and not unique to vdr) that is different such that the numbers being seen/displayed (and in particular "guest" numbers) now vs before the upgrade do not lend themselves to a direct comparison. That something is leading the numbers now to appear higher than they really are (at least in relative terms), or it could be that the numbers before were appearing lower than they really were. And by "something" I mean perhaps the way the numbers are gathered, the timing, or the source of information the numbers are derived from.
Again, my suspicions are not raised soley by what has happened here at vdr. It is this coupled with the fact that now when I go look at other sites (using similar board oftware) and see their numbers and really look at them, I see these glaringly high "guest" numbers (and also skewed relative the "member" numbers at the time). Before I just kind of accepted the "guest" numbers, but now...
Now a legitimate question would be "so what?". Well, the only so what is that if one is going to be analyzing the effects of the board upgrade in regard to online activity changes ya really have to be sure a) that the numbers are a meaningfull reflection of the number of people using or lurking at the board at a given time, and b) that the numbers now can be compared properly and directly with numbers from before (the upgrade).
Actually, having poked around a bit I think JP might have hit on something (at least a part of it) several posts back when he said "I assume bots account for some portion, since this version is far more spider friendly". It seems this has been a common suggestioned reason (bots and deep indexing, spiders, etc) given to people noticing big jumps in online users related to vbulletin 3. It does make sense at least in terms of skewing the numbers (which would come in in the "guest" figures) high. And if this is the case then it does mean that to compare online numbers (other than member counts) now to the pre-upgrade environment cannot be readily done as it is comparing apples and oranges (or perhaps as our friend Train says: apples to dump trucks :D ).
Addendum: One of the suggested features for upcoming versions of vbulletin I see is to have Who's Online show members + guests + bots as opposed to just members + guests as now.
Addendum 2: I see a post over at Daniweb where someone asked last year about "why all the guests". And they were told essentially it was caused largely by bots, etc.
Addendum3: Just say one post out and about where in response to a question about bots in the guest count one guy said about his experience "I had a bot crawling my forum early today. The number of forum guests were 200ish instead of usual 5ish." So it seems they can rack up some (false) numbers.
Until and unless we we determine the origins of the guest counts they do not mean to much statistically. So far we have two sources, true guests and search bot queries. And if they index us how many queries will we get? Gots to do some looking up here.
Absolutely GG. And, going beyond even statitistics one could say if a suficient number of these connections are not real people then things like the bandwidth impact implications and perhaps the notion of blocking some of these connections while letting others through could be worth looking at. I see reference to (I think it is) a file called robots.txt as a means of controlling things potentially.
And why the heck would we want to block search engines from indexing the site? We want the bots here, high search engine rankings are a good thing.
I fail to see the relevance at all of whether the number of guests online is 100% accurate or not, nor do I see any relevance in what proportion of guests are indexing bots. As a guide for comparing how well we're doing with other sites running the same software it has its uses, beyond that it is a fairly meaningless figure. The much more important metric of how many new members are joining (which is, of course, verifiable - BJ gets notified of every new member registering), has shown a marked increase since the upgrade. That increase is largely propotional to the increase in guests online, which suggests that the figure isn't too far out actually, and is exactly what I would expect. To join you first have to find the site, which is now possible since we are back in the search engine rankings again, then you have to be a guest before you can register. That is what has been happening since the upgrade and there isn't a thing that's suspicious about it.