Thanks to all of you for your patience regarding the technical aspects of the board, particularly the outages. Here’s a little summary of the situation.
Some weeks ago Rich and I noted that the board users were getting lots of errors. Discussed options and decided to try a couple of things. Rich did the software upgrade. Also moved the forum from one server on Dreamhost (our hosting company, at least for now) hoping that the relocation might help. Seemed like it did for awhile. But, we’ve had some rough days.
Rich has also been bowing out, although he continues to try to help.
I’m filing support requests regularly with Dreamhost and other than the general observation that the size of the database and the traffic on the board might be posing a problem, things haven’t gotten better.
One option under consideration is to move to a virtual private server option on Dreamhost. It looks to me like that would triple the monthly cost. Maybe it would take care of the problems. We could fund it through the Google ad revenues, which I’m scheduled to start getting soon. (Hey, I’ve never really been able to make money on C&F. Why start now?)
Another option is to research and move to another hosting service, in hopes that we can have a more robust forum without increasing the monthly fees. That’s time consuming but probably needs to be done.
If anybody out there is a real expert on this stuff, do let me know.
So there you go. I apologize for the problems in the interim.
Would it help at all to archive very old threads off in their own closed message board, for viewing only, and only have to manage relatively current activity in the active board?
Actually, Rich and I once talked about a purge of old posts. Frankly, there’s a large number that wouldn’t be missed and we developed a set of rules for eliminating some. And, we actually did delete a good number, although not as many as discussed. Deleted the old Political forum.
I probably asked this once on a previous search for new hosting: What are your requirements? Is there a spec sheet somewhere that could be sent to me (email or PM)? This would include what size the database is (indexes, text, etc.), and monthly bandwidth usage, current or projected, and any other special hosting requirements (software capabilities besides the phpBB).
As I understand it, the message board databases should only consist of text messages (wordy little b*stards aren’t we?), formatting codes, indexing, member profiles & avatars, PM boxes. All other posted images and mp3s, etc. are linked to off-site hosts at various locations (box.net, imageshack, etc.) and not managed by the message board. Such images should not contribute at all to your site’s bandwidth requirements, as they’re pulled in by the viewer’s browser via connections to those other sites.
BTW, any possibility that the most recent problems are due to the Internet outages due to the 3 cables cut near Egypt?
It’s bad enough that I have moved several clients away from them, despite the low cost, technically nice platform, and generous storage.
You specifically want to be looking for MySQL outages, and outages on the various clusters that host the MySQL dbs.
It might be worth looking for an alternate host. EApps offers some decent CentOS virtual hosting that would support this board. There are many others. But moving hosts is not easy or fun.
I would keep the informative posts on the whistle board, flute board, etc. These posts prevent the same things from being asked over and over and over (in theory at least), and are useful to keep around because they’re directly relevant to C&F’s main purpose.
Conversely, I would allow the posts on the Pub and the PROCTology forum to be deleted after a set period of time (a month, six months, a year, or whatever you decide) because most of them are not really relevant in the same way as the instrument and music ones are (especially the ones that change with the election cycles, seasons, weather, etc. and are then no longer relevant). This kind of move seems like it would free up a bunch of space while at the same time keep the good stuff where it is, no?
I belong to one message board where after a year, all threads that haven’t been posted in are deleted, except for sticky threads. And I’m pretty sure locked threads are automatically deleted after a certain amount of time, too. That board has never had problems like C&F has recently had. I don’t know if it’s related or not. My hunch is that it is, though.
I’m not too terribly well versed in these things, so if what I said doesn’t make sense, never mind.
Update: I’ve ordered an upgrade at the hosting service which should be in place in a day or so.
Wait 'til you guys get the bill.
In the meantime, I think resident genius moderator MTGuru may have come up with the idea that’s going to explain why the last 36 hours have been so bad, and hopefully this had led to a bit of an improvement.
I had asked Rich to help et up the RSS feed and he did and sent me the directions for setting it up and activating it. Since I did, we’ve had much worse problems that normal with the functioning of the board. That’s deleted (I think) as of 10 minutes ago, so I’m hoping that will make a difference.
Meanwhile, let me add that there’s something everyone can do that might help while we’re waiting for the upgrade to take effect.
Some of you are probably used to keeping multiple browser windows or tabs open on the board while you’re reading. But each window/tab may use up a bit of additional memory on the server. And right now, shared server memory is at a premium.
So if you can keep the number of your C&F windows and tabs to a minimum for now, and log out and close your windows/tabs when you’re done, that might help to allow everyone access.
That’s interesting. I thought that once a page was served,
the browser’s connection would be released. I’m surprised
it’s persistent.
Have you guys considered a tip jar? I’d be willing to pony
up for continued uptime. Of course, I guess that opens you
up for “I donated good money, and I still get 404’d. Waaaaa!”
I’ve been told the same thing by a very knowledgeable friend. The thing is, our hosting service seems to use a resource restriction protocol based on a number of things - including, apparently, IP hits, pages served, etc. So the idea was to keep the hit count lower for now and hope for better performance.
But it’s obviously not working, and we honestly have no idea what’s going on. IMO, there’s no reason a minimally robust server, even shared server, shouldn’t be able to handle the fairly simple requirements of the Chiffboard. If resources are the issue, then VPS should give us instant relief. If not, we’re preparing alternatives. Believe me, Dale and the mods are as tired of dealing with this as anybody.
If the alternatives are going to be costly, and if ads aren’t covering the costs of maintaining C&F, what would be a reasonable monthly or yearly membership fee? I ask because I have no idea of the costs involved. I would not want to put too onerous a burden on Dale, and if we can all chip in a moderate amount to keep things running smoothly I think it would only be fair.
The ad revenue should cover the new costs. If I ever need to, I’d just put a Paypal Donate button up some place. I’m good. Very nice of you, djm, to ask, though. Money’s not the problem. It’s logistics and the fact that all of this hit at Christmas and with some stuff that’s going on with me and my family.
I also guess that the problem lies in the web server configuration. Persistent HTTP connections are of course possible (and probably allowed here), but they can also be closed server-sided after a HTTP 200/OK has been sent. MySQL connections also can be persistent, but you have to open a persitent connection yourself (in the program/script), they normally are not persistent as a standard.
I doubt that you’ll be able/allowed to change the web server configuration if this is a standard web hosting service. You could ask, though. I ran a dedicated server with root access and all that stuff for two years, and it’s amazing to see how many tons of stuff can be configured. We ran loads of software on that machine and never ever brought it to it’s knees.
It’s all pretty frustrating to me. It would help to really know what the problem is. Dreamhost support and rich both think the board has outgrown Dreamhost’s regular service. Of course, Dreamhost has something to sell me. I’ve had five or six C&F members in touch with me…maybe 10 by now…and, interestingly, a whole range of theories about the problem and a range of ideas about what to do next. And, here’s me, with probably less knowledge than any of these folks about all of this, and so I’m pretty bewildered.
Intuitively, and that’s all it is, I think there’s a problem with configuration and I think if Dreamhost had adequate service, they could fix it, at least well enough that we could soldier on, giving me more time to seek some kind of permanent solution.
I’ve opted for the Dreamhost PS solution because they make the change for me and I don’t have to take on the task of trying to learn how to move the board and move it at this most inopportune time in my life.
Now I’m about to his SUBMIT and get an Internal Server Error message.
Well, since dealing with heavy loads on servers is part of my job, and with PHP and Apache, which is actually running this site (I think), I could give you my own thoughts. There are a couple of other techies here(and Rich) and I’m sure they told you the same… First of all, it has nothing to do with the Egypt line or whatever, it’s definitely the server. You wouldnt reach the ‘Internal server error’ if you werent reaching the site becausde of a cut cable
This isnt a database problem either, because you would probably get the famous ‘Mysql couldnt connect’ error, which happens when the traffic you have on your website is higher than what your database server can provide.
This is in fact an Apache web server error, which means the site seems to be too demanding for the web server. This can happen when the webserver is getting many requests (loading a page) and each request requires the web server to execute PHP scripts, which uses memory and CPU. Since the site is on a shared server, it’s probably overloading. (I don’t see how having 20 tabs at the same time can be worse than reloading the same page 20 times, the memory usage is going to be on YOUR computer for using too much memory because of many browser tabs, not on the web server).
What is suspicious is that instead of getting a clear message, you’re getting ‘Internal error’. It might be because the web server is actually running out of memory. If you were on a dedicated server, the server could be tweaked and optimized to run the site, but when you’re sharing, you can’t.
My own suggestion would be to shop for a dedicated server somewhere, if you can afford, or look at godaddy.com and try to see if what’s the best accounts they offer. There are a couple of bad things about godaddy.com, but they might be worth a try because they’re usually very cheap and configured properly.
second, you have no idea what else the server is doing
third, without access to the server logs there is no way to determine much of anything (past it ain’t working)
fourth, it’s a lot like the best starter whistle
it’s not likely that the geeks are gonna agree, especially without the logs
too many variables, shooting in the dark
oh, yeah…ya’d think that there would be a time of day when it work work well if it was a server too busy thing. I’ve been sickly & prone to hacking in the middle of the night. I doesn’t work any better then.
Agreed. At Dreamhost you are in a (massivly) shared web hosting environment. You get a slice of pie, but please don’t ask for a larger slice if needed. Dreamhost is a great value for a small website that does not require high availabilty or near 100% uptime. Likewise with GoDaddy, Bluehost and most similar services for under $25 a month.
Dedicated servers are getting really expensive. 2.5k-10k anually, not including administration. Rackspace.com is a good example. I’m not sure how that could be sustained? Even with donations… Perhaps someone with such a server is willing to donate space and bandwidth?
The middle ground are the virtual servers – basically fooling one computer into acting as many. They are more scaleable that the Dreamhost class of service, but not as much so as a dedicated server. Usually you can get hosting packages in the $150 a month range, with enough horsepower, memory, and bandwidth to sustain a popular site. eApps and WiredTree are two examples.
In my case, Dreamhost proved to be too unreliable for even a modestly popular site. We also had several hacking incidents, which I suspect are attributable to security problems at dreamhost (lots of users, lots of hackers). We moved away from them because of this.
One thing has changed with Dreamhost since my tenure there, in that they now offer virtual hosting. Perhaps that is what you are upgrading too. It may be an improvement. I do think that their tech support was fairly decent for an email only service, if you knew the right questions to ask.
attempt #10 - the server is dropping packets - it sux - pipe to small, processor brain-dead, technicians there for the doh . tried long technical and nice but got sick of losing text
