phpBB : Critical Error
Could not connect to the database
Getting this error about half the time when tring to access forums/threads/new post screen.
General maintenence, something I missed or is this a problem? It's a bit annoying, but annoying isn't all that bad.
Mages do it with Sheep.
Posts
Also got this lovely thing:
Link (much larger version)
It's that cursed pconnect Alpha is using, consarnit!
I told you! I told you all!
The world is ending.
Ennnnnnnnnnnnnnnnnnnnding!
Ahhhhhhhhhhhhhhhhh!!!!!!!!!!!!
:roll:
On the bright side, we're not double posting anymore.
Doc: That's right, twenty five years into the future. I've always dreamed on seeing the future, looking beyond my years, seeing the progress of mankind. I'll also be able to see who wins the next twenty-five world series.
Doc: That's right, twenty five years into the future. I've always dreamed on seeing the future, looking beyond my years, seeing the progress of mankind. I'll also be able to see who wins the next twenty-five world series.
All the current issues have been MySQL related. Alpha said he was gonna tinker with MySQL stuff.
Draw your own conclusions.
Just so I don't come off as a complete asshole: this is just speculation. I don't know exactly what he's working on, it's just a strange coincidence.
Things were fine this morning when the user load was low, but now that more people are on-line, the problem has gotten steadily worse and worse.
By that time everybody will have left the forum
Doc: That's right, twenty five years into the future. I've always dreamed on seeing the future, looking beyond my years, seeing the progress of mankind. I'll also be able to see who wins the next twenty-five world series.
Assmate. You did that on purpose. Funny, I got error when trying to open this thread
And when I tried to post this the first time.
Yes. My thought is that it's related to Alpha switching to persistent MySQL connections. All the connections are gobbled up by idle users and active users get screwed.
Assmate? As in you guys share an ass?
:shock:
It's a constipatery!
It started off happening about one in every ten pages I tried, now it's virtually every other page. However I am glad it's not my computer being an asshole.
GW2: Tavalisk.9463
I agree with this assessment.
The pconnect has been removed and the 250 hung threads have been painstakingly killed one-by-one. :!:
No ill-will toward alpha. He had no way of knowing this would happen.
Wouldn't that have been much easier?
(or something to that effect)
A lot of people could have let him know. I think this is justification for a Programming Forum. Not only could the forumers discuss programming, but if alpha or anyone has a question when it comes to optimizing the forums, he could ask it and get lots of feedback!
Suicide-squad, attack!
You may want to fill him in on some of the no-nos code wise.
I think what Ramius was saying was that no one could've expected it. I of course reserve the right to be wrong about this.
I hope that isn't what he's saying. Most people with experience with pconnect (there are at least a few) could have expected this. pconnect sets up a persistent connection that doesn't die. You cannot close pconnects, which is why there were 250 hung threads.
The best advise is to not use pconnect unless you need it, which isn't the case with the board.
I might have read into what Ramius said, but it seemed to me that he knew what the result of Alpha's coding would be.
At least two of us knew what was gonna happen.
I have used pconnect before, infact I currently am on another system and I was fully aware of the possibility of error, however, given the size of the board, it was an experiment I was willing to try as the potential gain was high. If you noticed I didnt just apply the patch, I made a thread for telling when pconnect broke, as I was aware of the risk and I stayed on for several hours waiting to see if there would be any problems.
Kakos, you mentioned me posting about it, I did. I recieved feedback from both sides, though I didnt see your name as one of them. I am always happy to talk about coding with people. I dont think that it merits spliting this forum in two however.
Appologies to Ramius, killing mySQL threads is no fun. Appologies to the forumers, because critical errors suck.
I know fuck all about the technical side of things, which is why I reserved the right to be wrong.
We have historically not used pconnect anywhere on the server. That is, for as long as I've been here it hasn't been used. I never bothered to give serious thought to why not because I never saw a noticable performance issue related to not using pconnect. I'm not saying I haven't seen performance problems on the server, just that if connecting to the database were slow, it would always be slow. And many times the server runs great.
I have a glint of a memory from a couple years ago, where we had problems with connections to the database. This is while I was still rather new to penny-arcade programming. And the message passed to me from HomeLAN was something like "you aren't using persistant connections, are you?"
We weren't, but that comment stuck with me as an indication that we shouldn't. If not for that comment way back when, I probably might have gone ahead and tried it at some point to see what kind of difference it might make. That is why I don't blame alpha for trying it.
The fact is, pconnect can work wonders in certain situations. I have used it in applications at work a number of times. But success or failure with using pconnect relies heavily on the webserver and database configuration, and here at HomeLAN, not only do we not have any control over those factors, but we also don't have much data about them. Anything I know about the server and network configuration of the cluster I know from empirical evidence acquired over time. I don't mean to disparage HomeLAN, I think they are awesome and have been immeasurably generous to penny-arcade. But they run this operation in such a way that there is virtually no information-exchange between them and me.
I typically wouldn't even share this kind of information with the forum-at-large. But I made a mistake in posting my earlier comment, because it cast some doubt on alphamonkey's abilities and/or judgement. That is a terrible "off-to-the-wrong-foot" way to transition from one administration to the next, so I want to set the record straight. Programming at penny-arcade presents some unique challenges, and I have full confidence in alphamonkey to meet those challenges. While there will be instances, like this one, where someone around the forums might say "I knew better" of a decision, alphamonkey should not be judged based upon a single decision. I believe alphamonkey is particularly well-suited for this position and I think that looking back 6 months from now you will find yourself feeling he is as dependable and ingenious as I ever was.
Hey man, no need to apologize. People don't find better ways of doing things without taking a risk and trying a new approach. With the exception of Ramius having to kill some connections, you didn't put anyone out, and I'm sure Ramius didn't mind all that much.
Honestly, I'd much rather have someone around that comes up with innovative ideas to better the forum. With Ramius leaving, I couldn't think of a better choice for him to make.
I had not seen the thread. I apologize. I also apologize if I came off as an ass. The information I had at the time was that you just applied the patch, which is not an unreasonable assumption with most forum admins. Thank you for actually starting a thread about this.
As far as experimentations on these forums, we could employ something that most development projects do. We have a 'stable' forum and an 'unstable' forum. You implement code on the 'unstable' forum and if it works, you put it in the 'stable' forum.
As far as testing the 'unstable' forum, you could ask for volunteers to use the 'unstable' forum. Or you could simply code up something that randomly redirects people to one or the other when they point their browser to http://www.penny-arcade.com/forums and can be turned off at a moment's notice.
EDIT : Its not your ISP, my guess would be you are either taking a very long time and letting your session expire or something very wrong is happening with your cookies.
Regarding the new threads/posts: that's due to the time difference among the servers in the cluster.
Regarding the login: Alpha was right. It's probably a cookie issue. Grease your browser's cookie(style) sheet and turn down the temperature on your firewall a bit.