Advanced search

Message boards : Server and website : scheduler and http errors

Author Message
Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 131 - Posted: 28 Jun 2007 | 23:00:14 UTC
Last modified: 4 Jul 2007 | 23:08:06 UTC

I\'ve been getting lots of http errors, I kind have been ignoring them as I thought it was my system, but I found out a possible cause (as reported a similar issue on the developers project list).

Bad source code by the developers, actually a typo.

you need to remove the \"space\" before </scheduler>

<!--<scheduler>http://www.ps3grid.net/PS3GRID_cgi/cgi </scheduler> -->

additioanlly your project is missing a link line

if you have this in the home page:

<!--
\";
include \'schedulers.txt\';
echo \"
-->
\";

change it to:

\";
include \'schedulers.txt\';
echo \"

then in schedulers.txt

<!-- <scheduler>http://www.ps3grid.net/PS3GRID_cgi/cgi</scheduler> -->
<link rel=\"boinc_scheduler\" href=\"http://www.ps3grid.net/PS3GRID_cgi/cgi\">

the link element is outside the comment.

The reason is some ISP\'s filter comments, thus thus participants client\'s can not contact the scheduler. They have both for compatibility with older clients.
____________
Alpha Tester ~~ BOINCin since 10-Apr-2004 (2.28) ~~~ Join team USA

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 146 - Posted: 4 Jul 2007 | 23:14:34 UTC

I still get constantly during uploads http errors, and then there is a long delay before the next attempt. Eventually the files get uploaded in about 13-24 attempts.

two questions:

#1 Does anybody else see these errors or is it just me ?

My satellite connection is great at downloads, like dsl speed, but poor at uploads, like dial up. and with 5-6 computers sharing the connection uploads are not so good sometimes.

#2 is there any way to compress the upload files or are they already compressed ?
The 23,000 KiB and 23,000 KiB are the ones giving me problems.

All my other computers on other boinc projects don\'t usually have problems, but the files are not quite as big, sometimes they take longer to upload, but without the http error ?

Log:
7/4/2007 6:45:33 PM Access to reference site succeeded - project servers may be temporarily down.
7/4/2007 6:45:28 PM Backing off 3 hr 13 min 13 sec on upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 6:45:28 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_0: http error
7/4/2007 6:45:28 PM Project communication failed: attempting access to reference site
7/4/2007 6:39:22 PM [file_xfer] Started upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 6:08:54 PM Access to reference site succeeded - project servers may be temporarily down.
7/4/2007 6:08:46 PM Backing off 2 hr 25 min 11 sec on upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 6:08:46 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_0: http error
7/4/2007 6:08:46 PM Project communication failed: attempting access to reference site
7/4/2007 6:02:11 PM [file_xfer] Started upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 5:56:38 PM Access to reference site succeeded - project servers may be temporarily down.
7/4/2007 5:56:32 PM Backing off 1 hr 49 min 12 sec on upload of file NQ16754-RUNCF-nodelete_0_4
7/4/2007 5:56:32 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_4: http error
7/4/2007 5:56:32 PM Project communication failed: attempting access to reference site
7/4/2007 5:50:22 PM [file_xfer] Started upload of file NQ16754-RUNCF-nodelete_0_4
7/4/2007 5:50:15 PM Access to reference site succeeded - project servers may be temporarily down.
7/4/2007 5:50:08 PM Backing off 1 hr 22 min 36 sec on upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 5:50:08 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_0: http error
7/4/2007 5:50:08 PM Project communication failed: attempting access to reference site
7/4/2007 5:42:24 PM [file_xfer] Started upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 5:38:50 PM Access to reference site succeeded - project servers may be temporarily down.
7/4/2007 5:38:42 PM Backing off 3 min 39 sec on upload of file NQ16754-RUNCF-nodelete_0_0
7/4/2007 5:38:42 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_0: http error
7/4/2007 5:38:42 PM Project communication failed: attempting access to reference site
7/4/2007 5:22:28 PM [file_xfer] Temporarily failed upload of NQ16754-RUNCF-nodelete_0_0: http error

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 147 - Posted: 5 Jul 2007 | 8:42:06 UTC - in response to Message 146.

[quote]I still get constantly during uploads http errors, and then there is a long delay before the next attempt. Eventually the files get uploaded in about 13-24 attempts.

two questions:

#1 Does anybody else see these errors or is it just me ?

I have never seen any problem in uploading, but again we are close.


#2 is there any way to compress the upload files or are they already compressed ?

Yes, we could compress this. Let me look into it.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 148 - Posted: 5 Jul 2007 | 8:43:43 UTC - in response to Message 146.

[quote]I still get constantly during uploads http errors, and then there is a long delay before the next attempt. Eventually the files get uploaded in about 13-24 attempts.

two questions:

However, if you had this problem yesterday then the reason is that the server was down for several hours.

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 149 - Posted: 5 Jul 2007 | 11:32:21 UTC - in response to Message 148.

However, if you had this problem yesterday then the reason is that the server was down for several hours.


Yes it was at the time of my post, but I had been and continued after that to hit, retry transfer. It would send a small chunk like 224KiB or 776KiB and then stall again. I would wait a few minutes and then retry.

And this has been happening almost every day on every upload on the PS3, but I do not watch every upload.

and being that I have only 1 PS3 attached to PS3GRID and nothing else it is hard to compare it against the other 5 computers attached to multiple projects. With those if I have a problem on 1 or all at the same time with a project its easy to figure out if it is just the 1 computer or the project.

That is why I am wondering if it is on my end or your end.

If no one else has this trouble it must be my end.

I will see if there is something I can do here, but I do not know linux that well. I do have another computer with Ubuntu linux on it and I\'ve tried to compare some of the settings between the 2, but even this is hard as YDL and Ubuntu do not have the exact same programs and layout.

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 156 - Posted: 8 Jul 2007 | 12:47:12 UTC

I turned on http_debug flag.

This appears to be a timeout error, when other projects on my connection upload, this interrupts PS3GRID\'s upload and it times out.

I emailed a copy of log to Gianni.

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 165 - Posted: 9 Jul 2007 | 11:29:22 UTC

I\'m still trying things on my end, no success yet.

I did set the PS3 client to only make 1 transfer at a time.

I tried setting http_1_0 which seemed to be better, but the files that uploaded early this morning still show a timeout during one transfer.

I was able to get 2 results from alpha test, 1 each per windowsXP computers, which is a 4,700KiB file to upload simultaneously and when a 3rd computer uploaded a small result for another project, there was not an interruption shown, they kept on transmitting. These are client version 5.10.13.

It keeps going back to a PS3 boinc client (or YDL on PS3) to PS3GRID issue.

The hard part in testing this is I cannot connect other computers to PS3GRID, nor can I connect boinc on PS3 to other projects as no other projects yet have officially supported applications. And waiting 22 hours between each test.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 173 - Posted: 9 Jul 2007 | 23:38:23 UTC - in response to Message 165.

Scheduler and timeout have been adjusted. Everything should be fine now.

Thanks for the suggestions.

GDF

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 180 - Posted: 11 Jul 2007 | 10:51:23 UTC

Thanks for the fixes.

This might have helped, I\'m not sure if my problem is solved yet.

I did get 60% of the 23,000 KiB file to go through at about 12.19KiBps until another host downloaded a task (10sec) and then uploaded a result (10sec) but the PS3 did not resume communications, the timeout was reached after 300sec.

Then the 3 smaller files (600, 600 and 0.04 KiB)went through OK.

Then an attempt on the 10,000 KiB file, it got about 1/2 done before another interruption from a diownload and upload.

After it restarted again I browsed here, which interrupted it too.

I can watch my modem and when this happens the transmit/receive lights will go off. So I know that the PS3 is not attempting communication, it (boinc ?) is just waiting for the timeout. This to me seems like a boinc thing.

While writing this message the 10,000 KiB file finished.

Then the 23,000 file resumed. if got about 1,800 more sent when other computer uploaded a result (40 sec), but this time it kept going and is still going....

6,999 KiB sent, only a little to go...

and then the other computer downloaded (10sec) from a different project this time,

but the PS3 finished.

So there were about 4 other possible interruptions, 2 did and 2 did not.
There were a lot of just scheduler requests from 2 other computers to several other projects and those seemed to have no effect. Email should have checked a couple of times during this and I think that was mostly no effect.

So I\'m a little confused <scratches head> if it is fixed or not.

Will have to wait for the next result to complete (about 16 hours).

It does seem to be better than it was.

Profile Krunchin-Keith [USA]
Avatar
Send message
Joined: 17 May 07
Posts: 512
Credit: 111,288,061
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 269 - Posted: 28 Jul 2007 | 21:00:43 UTC

With the change in my internet connection from Satellite to DSL my problem has gone away.

Last result uploaded without interruption or error. :)

The PS3 sent 2 files at the same time,

The 13,000KiB file took 7-8 minutes once it got started and the 23,000KiB file took about 12 minutes, but that was while the other files were uploading and downloading at the same time and I was browsing the internet on another computer.

So problem solved.

Post to thread

Message boards : Server and website : scheduler and http errors

//