# Suggestion: reduce the search timeout to 30 seconds **URL:** \ **Category:** About This Message Board\ **Created:** [September 15, 2017, 12:28pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397 "2017-09-15T12:28:26Z")\ **Posts on this page:** 11\ **Page:** 4
**Author:** ![Joey\_P](https://avatars.discourse-cdn.com/v4/letter/j/919ad9/32.png) [@Joey\_P](https://boards.straightdope.com/u/Joey_P)\ **Post date:** [September 19, 2017, 5:14pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/61 "2017-09-19T17:14:12Z")
> [@drad\_dog](#): > > I would suggest 3 searches in a 15 minute period. Let people make a typo and recalibrate their search if they need to! In the name of decency… No. That’s considerably worse, IMO. We’d prefer unlimited searches (like virtually every other message board). I understand what you’re trying to accomplish, but if you look at the average, you’re asking for 3 searches in 15 minutes when we currently have 7, or 12 in an hour when we currently have 30. If you wanted to go that route, it would have to be at least 5 (maybe 10) inside of 15 minutes. But, still, I don’t want my searches to be reliant and having to nail down search terms within X tries or get timed out. I’ll counter your suggesting with removing the timer altogether and if you’re looking for a compromise, it would be that we try it for a few days or a week or two and see what happens. ---
**Author:** ![drad\_dog](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@drad\_dog](https://boards.straightdope.com/u/drad_dog)\ **Post date:** [September 19, 2017, 5:47pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/62 "2017-09-19T17:47:24Z")
> [@Joey\_P](#): > > No. That’s considerably worse, IMO. We’d prefer unlimited searches (like virtually every other message board). I understand what you’re trying to accomplish, but if you look at the average, you’re asking for 3 searches in 15 minutes when we currently have 7, or 12 in an hour when we currently have 30. > > If you wanted to go that route, it would have to be at least 5 (maybe 10) inside of 15 minutes. But, still, I don’t want my searches to be reliant and having to nail down search terms within X tries or get timed out. > > I’ll counter your suggesting with removing the timer altogether and if you’re looking for a compromise, it would be that we try it for a few days or a week or two and see what happens. It seems like you are not going to get unlimited searches. It’s inherent in our situation. I’m with you in principle. But I just want to stop timing people out after one or two shots. It ignores the whole meaning of what a search is. It’s just…anti-intellectual. ---
**Author:** ![Ramira](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/ramira/32/344_2.png) [@Ramira](https://boards.straightdope.com/u/Ramira)\ **Post date:** [September 19, 2017, 6:20pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/63 "2017-09-19T18:20:05Z")
The door is closed, the perfection of all was achieved in the classical era and no dangerous innovation should be permitted. ---
**Author:** ![hajario](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/hajario/32/171_2.png) [@hajario](https://boards.straightdope.com/u/hajario)\ **Post date:** [September 19, 2017, 6:30pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/64 "2017-09-19T18:30:09Z")
> [@drad\_dog](#): > > I would suggest 3 searches in a 15 minute period. Let people make a typo and recalibrate their search if they need to! In the name of decency… That’s not possible with the board software. The only knob to turn is number of seconds of timeout. ---
**Author:** ![gnoitall](https://avatars.discourse-cdn.com/v4/letter/g/bb73d2/32.png) [@gnoitall](https://boards.straightdope.com/u/gnoitall)\ **Post date:** [September 19, 2017, 6:31pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/65 "2017-09-19T18:31:24Z")
> [@Ramira](#): > > The door is closed, the perfection of all was achieved in the classical era and no dangerous innovation should be permitted. That was almost poetic. Look. I’m an engineer. To me, this is an engineering problem. Has the system architecture been improved meaningfully since the conditions in which the search limitations were set up? If the board is running basically the same systems, there’s no reason to expect searches to work better. Most capacity-based performance problems don’t spontaneously resolve. The only way I could imagine the system status quo being sufficient is if board usership has dropped to the point that the outdated hardware can now handle it. Since I’m not involved in technical management of the board, I can’t say one way or the other, but I’d be surprised if it were so. ---
**Author:** ![hajario](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/hajario/32/171_2.png) [@hajario](https://boards.straightdope.com/u/hajario)\ **Post date:** [September 19, 2017, 6:42pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/66 "2017-09-19T18:42:53Z")
> [@gnoitall](#): > > That was almost poetic. > > Look. I’m an engineer. To me, this is an engineering problem. > > Has the system architecture been improved meaningfully since the conditions in which the search limitations were set up? If the board is running basically the same systems, there’s no reason to expect searches to work better. Most capacity-based performance problems don’t spontaneously resolve. > > The only way I could imagine the system status quo being sufficient is if board usership has dropped to the point that the outdated hardware can now handle it. Since I’m not involved in technical management of the board, I can’t say one way or the other, but I’d be surprised if it were so. I am also an engineer. We originally had no limit. We went to something like five minutes at some point and then later dropped to two minutes. 30 seconds has never been tried. ---
**Author:** ![cochrane](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/cochrane/32/10441_2.png) [@cochrane](https://boards.straightdope.com/u/cochrane)\ **Post date:** [September 19, 2017, 7:42pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/67 "2017-09-19T19:42:31Z")
> [@hajario](#): > > That’s not possible with the board software. The only knob to turn is number of seconds of timeout. Can we turn it to 11? 😃 Spinal Tap reference for those who don’t get the joke, but I suppose if I have to explain it, it wasn’t all that funny. ---
**Author:** ![Fiendish\_Astronaut](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/fiendish_astronaut/32/3295_2.png) [@Fiendish\_Astronaut](https://boards.straightdope.com/u/Fiendish_Astronaut)\ **Post date:** [September 19, 2017, 8:03pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/68 "2017-09-19T20:03:36Z")
Tapatalk allows me to refresh new posts searches with absolutely no delay between them ( it’s called Timeline). So I presume this doesn’t tax the board servers in the same way as a normal search does. Sent from my iPad using Tapatalk ---
**Author:** ![drad\_dog](https://avatars.discourse-cdn.com/v4/letter/d/aeb1de/32.png) [@drad\_dog](https://boards.straightdope.com/u/drad_dog)\ **Post date:** [September 19, 2017, 10:01pm UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/69 "2017-09-19T22:01:32Z")
> [@Fiendish\_Astronaut](#): > > Tapatalk allows me to refresh new posts searches with absolutely no delay between them ( it’s called Timeline). So I presume this doesn’t tax the board servers in the same way as a normal search does. > Sent from my iPad using Tapatalk Maybe someone could explain the burdens the system is under that make this a problem, and why they do not affect other sites? ---
**Author:** ![engineer\_comp\_geek](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/engineer_comp_geek/32/504_2.png) [@engineer\_comp\_geek](https://boards.straightdope.com/u/engineer_comp_geek)\ **Post date:** [September 20, 2017, 1:59am UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/70 "2017-09-20T01:59:45Z")
> [@drad\_dog](#): > > Maybe someone could explain the burdens the system is under that make this a problem, and why they do not affect other sites? When you access a page here, you access a few records in the database. When you search, you access a lot more records in the database. It all contributes to overall server load. Repeated searches, like if you spell something wrong, really hammer the database hard, since it’s one massive batch of record accesses after another. Many years ago, we were having all kinds of server problems that were basically the result of server overload. You’d get things like pages timing out whenever you try to read a thread, or posts wouldn’t go through, all sorts of stuff like that. Since searches hammer the database pretty hard, adding a timer between searches was a quick and easy way to drop the load on the servers. It wasn’t enough though, and the problems were only really solved when we moved to bigger and better servers. The original timer was 5 minutes. After the move to the newer servers, the timer was dropped to 2 minutes. I can only assume that they lowered it instead of getting rid of it completely because even on the new servers, there was still an issue of server capacity vs. server load. As for why other sites don’t have the problem, that’s a simple tradeoff between user accesses and server capacity. You can run a web site on an off the shelf computer from Walmart, but once a lot of people start accessing it, the web site will quickly slow to a crawl and will become unusable. On the other extreme end of things, you can run a data farm like Google does and you’ll have so much capacity that millions of users can access the site simultaneously and they’ll all work just fine. At every point in between, there’s a certain amount of usage that any site can handle before it runs into problems. Sites that don’t have loading issues have more capacity than they need. Sites with loading issues have less than they need. Sites with occasional loading issues are right on the edge - they have what they need for most cases but not enough for the highest peak loads. So you might think the solution for us is simple. Move to a hosting site with a greater server capacity, or pay for more capacity if our current hosting site can handle it (many hosting sites are scalable). That’s fine, but how do you pay for it? The SDMB isn’t a public service. We are corporate owned and we’re expected to turn a profit. If we don’t make enough of a profit, then the site is no longer worth maintaining financially, and the SDMB disappears forever. One very important fact is that we are owned by a print media corporation, and print media has been dying due to internet information taking over. You compare us to other sites that are doing well, and yes, we have some issues. But you aren’t comparing us to our industry as a whole. Many message boards have disappeared forever. We’re still around. We’re doing a lot better than they are, because we aren’t gone yet. As I’ve said earlier, I don’t have access to the server logs and usage statistics, so I’m making a lot of guesses. But we have regularly had complaints about loading issues up until about a year ago. That’s a pretty strong indication to me that we don’t have a huge amount of server overhead available. We could move to a more powerful server, but the risk in that is if we end up costing more than we make in advertising, subscription fees, etc. then one day we’ll just disappear. With our current setup, we’re able to handle our current usage, and the cost vs. revenue is such that we’re able to keep the lights on. I don’t have any access at all to the financial side of things here, but I suspect that we aren’t rolling in cash so much that we could easily just toss a bunch of money at better servers. ETA: Searches on google (and tapatalk, I assume) don’t contribute to our server load because they cache their own copies of our posts and perform the searches on their servers instead of ours. ---
**Author:** ![Joey\_P](https://avatars.discourse-cdn.com/v4/letter/j/919ad9/32.png) [@Joey\_P](https://boards.straightdope.com/u/Joey_P)\ **Post date:** [September 20, 2017, 2:56am UTC](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397/71 "2017-09-20T02:56:54Z")
> [@engineer\_comp\_geek](#): > > the timeout was put into place to reduce the overall server load. As was already pointed out, hitting the New Posts button is a search, so there are more searches here than some folks realize. > > While the SDMB runs fairly smoothly most of the time these days, we still have occasional complaints about board performance. Admittedly, the complaints do seem to be getting less frequent, so perhaps it is time to consider reducing the timeout. However, while\*\* I personally don’t have access to server statistics\*\*, the fact that we still have occasional performance issues rather strongly indicates that we don’t have a huge amount of readily available server capacity. Reduce the search time, and it is fairly likely that we will see some more severe resource starvation issues, like pages failing to load. A long wait between searches is annoying, but pages failing to load completely breaks the board and that’s much, much worse. That’s the potential downside of reducing the search time - frequent timeouts just trying to access the board in general. > > There’s another issue that plagues some users. It’s what Siam Sam calls “zapping”. The board software seems to forget the last time that the user accessed the SDMB, so the New Posts feature breaks, posts in all forums show as read even when they haven’t been read, etc. This is not directly caused by server load. I don’t know what the issue actually is (because the IT folks never said in detail) but it’s something that can’t be easily fixed in the version of vBulletin that we are using. The problem affects some users more than others. Some don’t see it at all. One very important point that is relevant to this discussion is that the problem seems to get worse when the board seems to be suffering from other performance issues. > > Database errors and database crashes seem to be getting more common as well. This can also be related to server loading. > > I don’t know what all is going on behind the scenes right now. The re-skinning issues are pretty obvious since they are visible to everyone, but they are just a small part of what is happening. > > My opinion, **speaking as someone who does not have access to the server so I’m guessing** as much as the rest of you are, is that changing the timeout right now would be a rather risky thing to do. If it were my system (which it’s not), I would wait until the dust settles with everything else before even considering a change like this. > > Anyway, that’s my 2 cents. > [@engineer\_comp\_geek](#): > > Eh, if we get busy again (assuming we don’t change the search timeout), I’d expect the board to perform the way it did when it used to be busier. In other words, if traffic increases to what it was a few years ago, we’ll get about the same amount of complaints about board performance as we did then. > > It’s changing the search timeout that has the potential to make things go all wonky. Best case, the newer, faster servers we are on these days can handle the load, no biggie. Worst case, the board has massive problems, which was how it was when the timeout was first installed. During busy periods, pages would often fail to load, posts often wouldn’t go through, etc. > > Searches are resource intensive. You can’t allow more searches without increasing the server load. Past experience indicates that the server load from more frequent searches is significant. > > Like I said earlier,\*\* I don’t have access to server statistics so I can’t say how significant the impact would be\*\*, but it’s definitely going to negatively impact the overall server performance. Maybe a little, maybe a lot, maybe so little that nobody really notices, but it will definitely add to the server load. > [@engineer\_comp\_geek](#): > > Well, first of all, I’m not the one you need to convince. I have absolutely no control whatsoever over that timer. A lot of people seem to think there’s no down side to changing the timer. I’m just pointing out that potentially there is. > > I defend it because a lot of people seem to think the only reason the delay is still in place is because the IT staff here is a bunch of idiots. From my point of view, I see periodic complaints about **server problems that seem to be due to overloading, which (and again, I don’t have access to actual server statistics, so I’m guessing)** indicates to me that we don’t have any spare server overhead. There are times when this board is busier than others. When the board is at its worst, things don’t seem to work well. If we’re already having problems, then adding more load onto an already overloaded server is only going to make the problems worse. So I keep defending the timeout because it makes an awful lot of sense to me to keep it in. The folks that manage the servers aren’t the idiots that some folks here think they are. > > That said, we haven’t had too many complaints in the last year or so. Siam Sam’s zapping (which doesn’t only affect him, he’s just the one that reports the problem most reliably so I associate his name with it) still occurs, and while that apparently isn’t directly related to server loading, it does seem to get worse when the server is performing worse. If the subject had come up a year ago (and it probably did, it comes up often), there’s no way I would have agreed that it was time to change the timer. We had just experienced a spat of server problems not long before that. Now though, maybe it is time to try lowering the timer value. We’ve gone a decent amount of time with fewer problems. > > I’m not saying it’s not time to try it. I’m just pointing out that it’s not without risks, and that the timer has had a good reason for staying where it is, up until very recently. > > If it were up to me (and again, it’s not), there’s no way in hell I’d do it right now. They are making too many other changes. Let those settle out for a month or so, then change the timer. If we change it now and the board has loading issues, are the issues caused by the timer or are they caused by the other changes? Better to change things one at a time. That way if something breaks, you know what broke it. > > If it were up to me I’d also banish viglink to the fiery depths of Hell, but that’s something else that I have no control over whatsoever. > > As for me always telling folks how wonky the search engine is, let’s face it. The search engine is wonky. The search engine is tied into the version of vBulletin that we use, and as long as we continue to use this version, we’re stuck with our wonky search engine. If people complain about it, I can’t tell them how to fix it. The best that I can do is tell them that it’s broken (so they know the problem isn’t on their end) and point them to google because sometimes that works in cases where our search engine doesn’t. It’s the best that I can recommend under the circumstances, so that’s what I do. I’m trying to be as helpful as possible. > > Sorry if you don’t like me using the word wonky. I picked that word up from Stephen King many years ago and it’s stuck with me. I like the word. > [@engineer\_comp\_geek](#): > > That’s just silly. I’m not saying anything close to that. > > My best guess, based only on the frequency of board performance issues in the past, is that making a significant change to the timer like setting it to 1 minute or shorter will mostly work, but we’ll have a bunch of folks complaining about things not working right in periodic spats every month or so.\*\* But I don’t know if we are still running on the same servers that we were a year ago\*\* and earlier, and I don’t know what other changes are being made during this current upgrade other than the mobile skinning. > > But like I already said, if I were in charge of the timer (and I’m not), I would be willing to give it a shot at 30 seconds just to see what happens. I just wouldn’t do it while other changes are being made. > > vBulletin uses SQL. We have an admin but he’s not dedicated only to the SDMB. He has other duties as well. He is very responsive when we have issues, but I don’t know how much time he has to fiddle with things like tuning and optimization. > [@engineer\_comp\_geek](#): > > When you access a page here, you access a few records in the database. When you search, you access a lot more records in the database. It all contributes to overall server load. Repeated searches, like if you spell something wrong, really hammer the database hard, since it’s one massive batch of record accesses after another. > > Many years ago, we were having all kinds of server problems that were basically the result of server overload. You’d get things like pages timing out whenever you try to read a thread, or posts wouldn’t go through, all sorts of stuff like that. Since searches hammer the database pretty hard, adding a timer between searches was a quick and easy way to drop the load on the servers. It wasn’t enough though, and the problems were only really solved when we moved to bigger and better servers. > > The original timer was 5 minutes. After the move to the newer servers, the timer was dropped to 2 minutes.\*\* I can only assume\*\* that they lowered it instead of getting rid of it completely because even on the new servers, there was still an issue of server capacity vs. server load. > > As for why other sites don’t have the problem, that’s a simple tradeoff between user accesses and server capacity. You can run a web site on an off the shelf computer from Walmart, but once a lot of people start accessing it, the web site will quickly slow to a crawl and will become unusable. On the other extreme end of things, you can run a data farm like Google does and you’ll have so much capacity that millions of users can access the site simultaneously and they’ll all work just fine. At every point in between, there’s a certain amount of usage that any site can handle before it runs into problems. Sites that don’t have loading issues have more capacity than they need. Sites with loading issues have less than they need. Sites with occasional loading issues are right on the edge - they have what they need for most cases but not enough for the highest peak loads. > > So you might think the solution for us is simple. Move to a hosting site with a greater server capacity, or pay for more capacity if our current hosting site can handle it (many hosting sites are scalable). That’s fine, but how do you pay for it? The SDMB isn’t a public service. We are corporate owned and we’re expected to turn a profit. If we don’t make enough of a profit, then the site is no longer worth maintaining financially, and the SDMB disappears forever. One very important fact is that we are owned by a print media corporation, and print media has been dying due to internet information taking over. You compare us to other sites that are doing well, and yes, we have some issues. But you aren’t comparing us to our industry as a whole. Many message boards have disappeared forever. We’re still around. We’re doing a lot better than they are, because we aren’t gone yet. > > As I’ve said earlier, **I don’t have access to the server logs and usage statistics, so I’m making a lot of guesses**. But we have regularly had complaints about loading issues up until about a year ago. That’s a pretty strong indication to me that we don’t have a huge amount of server overhead available. We could move to a more powerful server, but the risk in that is if we end up costing more than we make in advertising, subscription fees, etc. then one day we’ll just disappear. > > With our current setup, we’re able to handle our current usage, and the cost vs. revenue is such that we’re able to keep the lights on. **I don’t have any access at all to the financial side of things here, but I suspect** that we aren’t rolling in cash so much that we could easily just toss a bunch of money at better servers. > > ETA: Searches on google (and tapatalk, I assume) don’t contribute to our server load because they cache their own copies of our posts and perform the searches on their servers instead of ours. That’s a lot of words to say “I don’t know”. I’m confused about your print media comment. You said " One very important fact is that we are owned by a print media corporation, and print media has been dying due to internet information taking over.", but didn’t really expand on that. If print media is dying out (and it is) and internet information is taking over, why do they continue to serve up malware infested ads cripple the software? Shouldn’t they do what they drive more people to the boards to generate more income to whoever owns us at the moment? If someone clicks to the SD homepage and finds their way to the message boards, pop ups and unders and their virus software going off isn’t a great way to keep them around (or get them to buy a subscription). Also, as for SD’s staying power, yes it’s been around for nearly 20 years in it’s current interation, but so have other boards. I’ve been a member (active and otherwise) on Tivocommunity and dslreports, both having been around for quite a while as well. DSLreports has been in existence since 1999. [Previous page](https://boards.straightdope.com/t/suggestion-reduce-the-search-timeout-to-30-seconds/796397.md?page=3)