digitalmars.D - DSource
- Kyle Furlong (8/8) Apr 25 2006 Okay, so we just got dug, and DSource has been down for more than a
- Walter Bright (2/11) Apr 25 2006 digitalmars.com saw roughly a 6 fold increase in traffic yesterday.
- Kyle Furlong (5/17) Apr 25 2006 Thats exciting, but what does that have to do with the fact that dsource...
- Lars Ivar Igesund (6/24) Apr 25 2006 The problems Dsource is having is not really traffic related (although h...
- =?ISO-8859-1?Q?Jari-Matti_M=E4kel=E4?= (4/9) Apr 25 2006 Has this already been fixed in the 0.10dev version of trac?
- Lars Ivar Igesund (5/15) Apr 25 2006 AFAIK, DSource is one of very few Trac enabled sites using Postgres as i...
- =?UTF-8?B?SmFyaS1NYXR0aSBNw6RrZWzDpA==?= (6/10) Apr 25 2006 Sad to hear. Postgres might be more elegant, but I use sqlite myself for
- Brad Anderson (18/21) Apr 25 2006 I was getting ready for Digg-type traffic, and I have seen Trac db's sto...
- Kyle Furlong (3/29) Apr 25 2006 I guess I got unlucky in that every time I looked in on the site in the
- Walter Bright (2/4) Apr 25 2006 That's fixed now.
- Brad Anderson (2/8) Apr 26 2006 sweet, thanks
- Kyle Furlong (6/30) Apr 25 2006 I understand all this. I never thought that dsource was down due to the
- Hasan Aljudy (3/12) Apr 25 2006 Has it been down for a week? I remember navigating dsource a few days
- Lars Ivar Igesund (7/20) Apr 25 2006 No, it hasn't been down at all, in the correct way of speech, but the Tr...
- lightoze (2/8) Apr 25 2006 May be this can be temporary solved by setting low connection timeout......
- clayasaurus (2/13) Apr 25 2006 or use the SQLite backend?
- James Dunne (12/33) Apr 26 2006 In my experience, SQLite has huge problems with concurrency. I've tried...
Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.
Apr 25 2006
Kyle Furlong wrote:Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.digitalmars.com saw roughly a 6 fold increase in traffic yesterday.
Apr 25 2006
Walter Bright wrote:Kyle Furlong wrote:Thats exciting, but what does that have to do with the fact that dsource has been down for much longer than just yesterday? It would have been nice to have some of that increased traffic funneled to a working dsource. What we got is in essence a PR disaster.Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.digitalmars.com saw roughly a 6 fold increase in traffic yesterday.
Apr 25 2006
Kyle Furlong wrote:Walter Bright wrote:The problems Dsource is having is not really traffic related (although high traffic don't help), it is a technical problem in Trac wherein it don't release it's db connections the way it should. You might help resolve the problem by finding out where in the Trac/Python code the problem is, I know Brad would be very happy to find it.Kyle Furlong wrote:Thats exciting, but what does that have to do with the fact that dsource has been down for much longer than just yesterday? It would have been nice to have some of that increased traffic funneled to a working dsource. What we got is in essence a PR disaster.Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.digitalmars.com saw roughly a 6 fold increase in traffic yesterday.
Apr 25 2006
Lars Ivar Igesund wrote:The problems Dsource is having is not really traffic related (although high traffic don't help), it is a technical problem in Trac wherein it don't release it's db connections the way it should. You might help resolve the problem by finding out where in the Trac/Python code the problem is, I know Brad would be very happy to find it.Has this already been fixed in the 0.10dev version of trac? -- Jari-Matti
Apr 25 2006
Jari-Matti Mäkelä wrote:Lars Ivar Igesund wrote:AFAIK, DSource is one of very few Trac enabled sites using Postgres as it's DB backend. Apparently there are no similar problems with the default SQLite backend. Therefore it has been somewhat difficult getting those who know the code best to look at it.The problems Dsource is having is not really traffic related (although high traffic don't help), it is a technical problem in Trac wherein it don't release it's db connections the way it should. You might help resolve the problem by finding out where in the Trac/Python code the problem is, I know Brad would be very happy to find it.Has this already been fixed in the 0.10dev version of trac?
Apr 25 2006
Lars Ivar Igesund wrote:AFAIK, DSource is one of very few Trac enabled sites using Postgres as it's DB backend. Apparently there are no similar problems with the default SQLite backend. Therefore it has been somewhat difficult getting those who know the code best to look at it.Sad to hear. Postgres might be more elegant, but I use sqlite myself for various reasons. It's very easy to dump the db, for example. Is there a good reason to use Postgres in the first place? -- Jari-Matti
Apr 25 2006
Jari-Matti Mäkelä wrote:Is there a good reason to use Postgres in the first place?I was getting ready for Digg-type traffic, and I have seen Trac db's stored in SQLite lock during write (yes, the whole db locks). So I chose Postgres. I have heard of using one database and separating the Trac tables for a project into different schemas - the problem that was observed separate of dsource's actually went away doing this. This is a big task and I don't have a lot of time right now. A few weeks ago when I had some time, the problem wasn't as pronounced... Murphy. I have tried some settings in postgresql.conf to limit connection time-outs and they're not working yet. Plus, I didn't see nearly the traffic Walter talks about getting on his site, because the link to dsource is reasonably obscure. Some of that lack of traffic was the brief periods when we were down, but I would have seen hits in the logs, just with errors. Also, we weren't down for an entire week as the OP suggests. Apache resets itself, so we were down no longer than 2 hours at a time, and then were fine for most likely 12 hrs, until it happened again. BA
Apr 25 2006
Brad Anderson wrote:Jari-Matti Mäkelä wrote:I guess I got unlucky in that every time I looked in on the site in the past week it was down.Is there a good reason to use Postgres in the first place?I was getting ready for Digg-type traffic, and I have seen Trac db's stored in SQLite lock during write (yes, the whole db locks). So I chose Postgres. I have heard of using one database and separating the Trac tables for a project into different schemas - the problem that was observed separate of dsource's actually went away doing this. This is a big task and I don't have a lot of time right now. A few weeks ago when I had some time, the problem wasn't as pronounced... Murphy. I have tried some settings in postgresql.conf to limit connection time-outs and they're not working yet. Plus, I didn't see nearly the traffic Walter talks about getting on his site, because the link to dsource is reasonably obscure. Some of that lack of traffic was the brief periods when we were down, but I would have seen hits in the logs, just with errors. Also, we weren't down for an entire week as the OP suggests. Apache resets itself, so we were down no longer than 2 hours at a time, and then were fine for most likely 12 hrs, until it happened again. BA
Apr 25 2006
Brad Anderson wrote:Plus, I didn't see nearly the traffic Walter talks about getting on his site, because the link to dsource is reasonably obscure.That's fixed now.
Apr 25 2006
Walter Bright wrote:Brad Anderson wrote:sweet, thanksPlus, I didn't see nearly the traffic Walter talks about getting on his site, because the link to dsource is reasonably obscure.That's fixed now.
Apr 26 2006
Lars Ivar Igesund wrote:Kyle Furlong wrote:I understand all this. I never thought that dsource was down due to the digg effect. I wish that had been the case! At least some of them would have seen the site. I was well aware of the connections issue and the length of down time which had been going on previous to the whole digg thing.Walter Bright wrote:The problems Dsource is having is not really traffic related (although high traffic don't help), it is a technical problem in Trac wherein it don't release it's db connections the way it should. You might help resolve the problem by finding out where in the Trac/Python code the problem is, I know Brad would be very happy to find it.Kyle Furlong wrote:Thats exciting, but what does that have to do with the fact that dsource has been down for much longer than just yesterday? It would have been nice to have some of that increased traffic funneled to a working dsource. What we got is in essence a PR disaster.Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.digitalmars.com saw roughly a 6 fold increase in traffic yesterday.
Apr 25 2006
Kyle Furlong wrote:Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.Has it been down for a week? I remember navigating dsource a few days ago and it was working!
Apr 25 2006
Hasan Aljudy wrote:Kyle Furlong wrote:No, it hasn't been down at all, in the correct way of speech, but the Trac part of the site that use the underlying database, causes connection leaks that makes those pages inaccesible until they are cleaned up. At that point it also affects the forums. The connections also cause some thrashing, meaning that svn may or may not be somewhat accesible too, but in general, svn has been mostly unaffected.Okay, so we just got dug, and DSource has been down for more than a week. Does anyone else see the problem here? People will be turning away as I write this because the main repository of open source development in D has a "FATAL: connection limit exceeded for non-superusers" message on the front page. While I appreciate the work done to create and maintain dsource.org, a week is not acceptable. I would be happy to help in any capacity to get even a mirror up asap.Has it been down for a week? I remember navigating dsource a few days ago and it was working!
Apr 25 2006
Lars Ivar Igesund says...No, it hasn't been down at all, in the correct way of speech, but the Trac part of the site that use the underlying database, causes connection leaks that makes those pages inaccesible until they are cleaned up. At that point it also affects the forums. The connections also cause some thrashing, meaning that svn may or may not be somewhat accesible too, but in general, svn has been mostly unaffected.May be this can be temporary solved by setting low connection timeout...?
Apr 25 2006
lightoze wrote:Lars Ivar Igesund says...or use the SQLite backend?No, it hasn't been down at all, in the correct way of speech, but the Trac part of the site that use the underlying database, causes connection leaks that makes those pages inaccesible until they are cleaned up. At that point it also affects the forums. The connections also cause some thrashing, meaning that svn may or may not be somewhat accesible too, but in general, svn has been mostly unaffected.May be this can be temporary solved by setting low connection timeout...?
Apr 25 2006
clayasaurus wrote:lightoze wrote:In my experience, SQLite has huge problems with concurrency. I've tried developing for it and it's really difficult to get multithreaded database access code to work. -- -----BEGIN GEEK CODE BLOCK----- Version: 3.1 GCS/MU/S d-pu s:+ a-->? C++++$ UL+++ P--- L+++ !E W-- N++ o? K? w--- O M-- V? PS PE Y+ PGP- t+ 5 X+ !R tv-->!tv b- DI++(+) D++ G e++>e h>--->++ r+++ y+++ ------END GEEK CODE BLOCK------ James DunneLars Ivar Igesund says...or use the SQLite backend?No, it hasn't been down at all, in the correct way of speech, but the Trac part of the site that use the underlying database, causes connection leaks that makes those pages inaccesible until they are cleaned up. At that point it also affects the forums. The connections also cause some thrashing, meaning that svn may or may not be somewhat accesible too, but in general, svn has been mostly unaffected.May be this can be temporary solved by setting low connection timeout...?
Apr 26 2006