Mongrel users - May 2007 - Problems with mongrel_cluster 1.0.1.1 on FreeBSD

I''m running FreeBSD 6.1 and installed the 1.0.1.1 mongrel_cluster
prerelease.  I had a problem with the mongrels not stopping, and it
turns out it''s because ps is reporting my processes as just
[ruby18]

instead of someting more like "mongrel_rails start -d ......."

Not sure why it''s doing that, but it is.  Any ideas?

For the time being I just changed line 165 of init.rb to read
pid = ps_output =~ /mongrel_rails|ruby18/ ? pid : nil

and that''s working fine.

Pat

On 5/4/07, Pat Maddox <pergesu at gmail.com>
wrote:
> I''m running FreeBSD 6.1 and installed the 1.0.1.1 mongrel_cluster
> prerelease.  I had a problem with the mongrels not stopping, and it
> turns out it''s because ps is reporting my processes as just
> [ruby18]
>
> instead of someting more like "mongrel_rails start -d ......."
>
> Not sure why it''s doing that, but it is.  Any ideas?
>
> For the time being I just changed line 165 of init.rb to read
> pid = ps_output =~ /mongrel_rails|ruby18/ ? pid : nil
>
> and that''s working fine.
>
> Pat
>
I''ve got two sets of mongrels running now, one running a staging
environment and the other production.

[twist at jack ~/twist/shared/log]$ ps ax | grep ruby
88908  ??  S      0:01.96 [ruby18]
88911  ??  S      0:02.15 [ruby18]
88914  ??  S      0:02.14 [ruby18]
89021  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
start -d -e production -c
89026  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
start -d -e production -c
89029  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
start -d -e production -c

The [ruby18] processes are all staging mongrels.  Any clue as to why
they wouldn''t show the full command?

Pat

I too am running mongrel on 6.1 but I have a few on 4.x as well.  I 
experienced problems with stopping mongrels UNTIL I switched the pidfile 
location in the config file to point to

/var/tmp/[site].pid

If you are using a relative path to your pid files in the config file, 
as I was when I was having problems, it wouldn''t properly stop. 
I''m not
sure of your exact circumstances but you might try switching the pidfile 
location.  This made ALL the difference.  I have NO problems with 
processes that won''t stop.

Using the /var/tmp directory is regarded as "best practice".  This is,
I
believe, mentioned in the release announcement for the 1.0.1.1 
pre-release.  Search the archives and you should find it.

matte

Pat Maddox wrote:
> On 5/4/07, Pat Maddox <pergesu at gmail.com> wrote:
>   
>> I''m running FreeBSD 6.1 and installed the 1.0.1.1
mongrel_cluster
>> prerelease.  I had a problem with the mongrels not stopping, and it
>> turns out it''s because ps is reporting my processes as just
>> [ruby18]
>>
>> instead of someting more like "mongrel_rails start -d
......."
>>
>> Not sure why it''s doing that, but it is.  Any ideas?
>>
>> For the time being I just changed line 165 of init.rb to read
>> pid = ps_output =~ /mongrel_rails|ruby18/ ? pid : nil
>>
>> and that''s working fine.
>>
>> Pat
>>
>>     
>
> I''ve got two sets of mongrels running now, one running a staging
> environment and the other production.
>
> [twist at jack ~/twist/shared/log]$ ps ax | grep ruby
> 88908  ??  S      0:01.96 [ruby18]
> 88911  ??  S      0:02.15 [ruby18]
> 88914  ??  S      0:02.14 [ruby18]
> 89021  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
> start -d -e production -c
> 89026  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
> start -d -e production -c
> 89029  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
> start -d -e production -c
>
> The [ruby18] processes are all staging mongrels.  Any clue as to why
> they wouldn''t show the full command?
>
> Pat
> _______________________________________________
> Mongrel-users mailing list
> Mongrel-users at rubyforge.org
> http://rubyforge.org/mailman/listinfo/mongrel-users
>   
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://rubyforge.org/pipermail/mongrel-users/attachments/20070504/9753ccbc/attachment.html

* Pat Maddox (pergesu at gmail.com) wrote:
> [twist at jack ~/twist/shared/log]$ ps ax | grep ruby
> 88908  ??  S      0:01.96 [ruby18]
..
> 89029  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
> start -d -e production -c
> 
> The [ruby18] processes are all staging mongrels.  Any clue as to why
> they wouldn''t show the full command?
They''ve probably been swapped out; -f can bring the full command name
back if you''re root.

-- 
Thomas ''Freaky'' Hurst
    http://hur.st/

On 5/5/07, Thomas Hurst <tom at hur.st> wrote:
> * Pat Maddox (pergesu at gmail.com) wrote:
>
> > [twist at jack ~/twist/shared/log]$ ps ax | grep ruby
> > 88908  ??  S      0:01.96 [ruby18]
> ..
> > 89029  ??  S      0:01.95 /usr/local/bin/ruby18 /bin/mongrel_rails
> > start -d -e production -c
> >
> > The [ruby18] processes are all staging mongrels.  Any clue as to why
> > they wouldn''t show the full command?
>
> They''ve probably been swapped out; -f can bring the full command
name
> back if you''re root.
What do you mean swapped out?

Pat

* Pat Maddox (pergesu at gmail.com) wrote:

(Sorry to the moderator(s), who I keep spamming by sending from the wrong
address.  I''ve added a folder-hook to prevent it)
> > They''ve probably been swapped out; -f can bring the full
command name
> > back if you''re root.
> 
> What do you mean swapped out?
I mean, the pages which contain the command line are not in system
memory, so it''s just showing you the command name instead of swapping
them back in, which could get quite expensive on a loaded system with
lots of inactive processes (think: 4k * 1000 processes for something the
user probably doesn''t even want).

You can turn it off by rebuilding it without the -DLAZY_PS line, ala:

cd /usr/src/bin/ps && sed -i .orig -e ''s/-DLAZY_PS//''
Makefile && make clean all install

But obviously this isn''t something mongrel should be depending on ;)

Maybe we could use a proper interface to process lists, with backends to
ps(1), procfs(5), kvm(3), Win32, etc.  I have a decent chunk of code for
FreeBSD/kvm(3) right here, though it''s not yet 64bit clean.

-- 
Thomas ''Freaky'' Hurst
    http://hur.st/