#290 Long-running foreground jobs

[Anonymous]

(it I understand the situation correctly) If I run the project-file and it takes longer than ~3 minutes (for example, if I made a mistake in C program and it led to an infinite loop), then all the wiki page freezes with the final message about SQL database error. While it would be nicer just to stop that project with the message

[Anonymous]

(ps: it was Andrei A.)

[Lee Worden]

Thanks Andrei! Would you be able to recreate this (in a browser tab you
aren't using, maybe) and record how long it takes to fail, and capture
the text of the SQL error message? I'm not familiar with this problem
before now. (I think it's probably related to the new server
configuration we're trying out.)

[Andrei Akhmetzhanov]

Yes, I'll try to do it. For me, it was not the first time with such message at the end

Thanks!

[Jonathan Dushoff]

I did this in the simplest possible way at http://lalashan.mcmaster.ca/theobio/projects/index.php?title=Yes. It took about 4 minutes to fail. The message was:

Bad Gateway!

The proxy server received an invalid response from an upstream server.

If you think this is a server error, please contact the webmaster.
Error (none)
lalashan.mcmaster.ca
Apache

[Lee Worden]

Thanks! I'm interested in knowing whether the SQL message Andrei
mentioned is something different from this. This looks like the proxy
at yushan is set to timeout after about 4 minutes of waiting for the
wiki software to return from whichever node. Since it can take 3
minutes, or whatever the time limit is set to in the php config, PER
PROJECT FILE, the timeout should be longer than 4 minutes.

[Andrei Akhmetzhanov]

Yes, Lee, I think I understood. I've got also Bad Gateway, this means that it's good.

Most probably, I got "SQL-something error" because I tried to make changes during those five minutes. Sometimes you don't know what's exactly happening and you try to refresh the page, to make some changes in the WW-page or clear the working directory.

That's good and sorry for disturbing.

Thanks for trying to help!

[Lee Worden]

On 04/23/2013 03:09 PM, Andrei Akhmetzhanov wrote:

Yes, Lee, I think I understood. I've got also Bad Gateway, this means
that it's good.

OK, I think Bad Gateway is due to a timeout setting on the proxy that's
distributing requests for wiki pages across the yushan cluster. I've
mentioned this to Jonathan, so hopefully it's being taken care of.

Most probably, I got "SQL-something error" because I tried to make
changes during those five minutes. Sometimes you don't know what's
exactly happening and you try to refresh the page, to make some changes
in the WW-page or clear the working directory.

It's very hard to guess what causes what. Best to have as much precise
data as possible so I can get my hands dirty in the log files and find
out what's really happening. If you do get an SQL error, please capture
the full text and tell me what happened, if possible.

Thanks for your help!

[Andrei Akhmetzhanov]

Yes, sure

Update: if one clears the working directory or try to change the text of source-files during freezing the WW-page and while an incorrect foreground job is running, "forever" jobs might appear on the server.

[Lee Worden]

"Forever" jobs are documented at https://sourceforge.net/p/workingwiki/bugs/143/. I have a question for you that I'll post there.