We're troubleshooting a similar problem to what Michael is
seeing. I'd really appreciate some suggestions on where to
look--we're hurting and need to figure this out.
We recently upgraded our Slash site to a newer faster box, and in the
process upgraded MySQL to the latest. Now we've had several instances
of the scheduled tasks grinding to a halt as the result of a process
that gets wedged.
In Michael's mail he listed an nidex.pl task that was running. In our
case, we've had both schedule index.pl and article.pl (out of freshenup.pl)
get stuck. These should normally take a second or so to process:
Fri Jan 24 16:55:35 2003 Updating 03/01/24/1339257
Fri Jan 24 16:55:36 2003 article.pl virtual_user=VUser ssi=yes sid='03/01/24/1\
339257' section='articles' bytes=10577
Also, like Michael, the last thing in our log was the 'Updating
yy/mm/dd/article' message from freshenup.pl. Since this process
didn't complete, no other scheduled tasks were started. Our
slashboxes weren't updated, and the article counts on the front page
weren't updating, etc. Basically the same symptoms that Michael's
reported.
We've studied the code extensively and concluded that the freshenup.pl
task is sitting at POSIX:waitpid(), waiting for article.pl. No clue
as to why article.pl is getting stuck--it mostly makes database calls.
When I kill the stuck task and rerun it manually (by invoking perl
article.pl etc) it always comletes.
Our operating environment is as follows. We upgraded some packages
during our move to a new box and suspect there's some interaction
between them, so I'm half expecting a "Slash is known to not work
with that version" response.
RedHat 7.3 base
Slash 2.2.6
perl-5.6.1-34.99.6
mod_perl-1.26-5
apache-1.3.27-2
mysql-3.23.54a-3.72
perl-DBI-1.21-1
Any suggestions are appreciated, including debugging hints.
As a secondary problem--not sure this is related--we've had the slash
process die with the following sql error twice now. This error we
haven't had time to fully investigate yet.
Tue Feb 4 06:07:00 2003 Send Admin Mail Begin
DBD::mysql::st execute failed: Got error 127 from table handler at/usr/lib/perl5/site_perl/5.6.1/i386-linux/Slash/DB/Utility.pm line 260.
Error in library:Slash::DB::Static::MySQL:/usr/lib/perl5/si te_perl/5.6.1/i386-linux/Slash/DB/Static/MySQL.pm: 329:SELECT count(*) FROM accesslog WHERE to_days(now()) - to_days(ts)=1 GROUP BY host_addr
Which was called by:main:/common/slash/runtime/site/msnews.bostoncu re.org/tasks/adminmail.pl:28:SELECT count(*) FROM accesslog WHERE to_days(now()) - to_days(ts)=1 GROUP BY host_addr
Can't call method "rows" on an undefined value at/usr/lib/perl5/site_perl/5.6.1/i386-linux/Slash/DB/Static/MySQL.pm line 331.
Thanks in advance,
Brian Del Vecchio
MSNews: http://msnews.bostoncure.org/
I'm running almost exactly the same versions on one of my boxes. Though I don't (currently) have Slash running on it, I didn't have the problem you're seeing when I was running it there. The only difference I see is that I have mysql-3.23.54a-3.73, not mysql-3.23.54a-3.72. I haven't taken the time to see if there's any significant difference between the version built for 7.2. and the one built for 7.3, but I'd look into this.
Stuck process (Score:1)
We recently upgraded our Slash site to a newer faster box, and in the process upgraded MySQL to the latest. Now we've had several instances of the scheduled tasks grinding to a halt as the result of a process that gets wedged.
In Michael's mail he listed an nidex.pl task that was running. In our case, we've had both schedule index.pl and article.pl (out of freshenup.pl) get stuck. These should normally take a second or so to process:
Fri Jan 24 16:55:35 2003 Updating 03/01/24/1339257 Fri Jan 24 16:55:36 2003 article.pl virtual_user=VUser ssi=yes sid='03/01/24/1\ 339257' section='articles' bytes=10577 Also, like Michael, the last thing in our log was the 'Updating yy/mm/dd/article' message from freshenup.pl. Since this process didn't complete, no other scheduled tasks were started. Our slashboxes weren't updated, and the article counts on the front page weren't updating, etc. Basically the same symptoms that Michael's reported.
We've studied the code extensively and concluded that the freshenup.pl task is sitting at POSIX:waitpid(), waiting for article.pl. No clue as to why article.pl is getting stuck--it mostly makes database calls.
When I kill the stuck task and rerun it manually (by invoking perl article.pl etc) it always comletes.
Our operating environment is as follows. We upgraded some packages during our move to a new box and suspect there's some interaction between them, so I'm half expecting a "Slash is known to not work with that version" response.
RedHat 7.3 base
Slash 2.2.6
perl-5.6.1-34.99.6
mod_perl-1.26-5
apache-1.3.27-2
mysql-3.23.54a-3.72
perl-DBI-1.21-1
Any suggestions are appreciated, including debugging hints.
As a secondary problem--not sure this is related--we've had the slash process die with the following sql error twice now. This error we haven't had time to fully investigate yet.
Tue Feb 4 06:07:00 2003 Send Admin Mail Begin DBD::mysql::st execute failed: Got error 127 from table handler at
Thanks in advance,
Brian Del Vecchio
MSNews: http://msnews.bostoncure.org/
Re:Stuck process (Score:1)
I'm running almost exactly the same versions on one of my boxes. Though I don't (currently) have Slash running on it, I didn't have the problem you're seeing when I was running it there. The only difference I see is that I have mysql-3.23.54a-3.73, not mysql-3.23.54a-3.72. I haven't taken the time to see if there's any significant difference between the version built for 7.2. and the one built for 7.3, but I'd look into this.
Good luck!
* [wdr1.com] uptime: 00:13:01 up 55 days, 13:18, 9 users, load
Parent