Re: unexpected Apache::VMonitor output

[prev] [thread] [next] [Date index for 2005/01/22]

At 05:30 PM 1/21/2005 -0500, Stas Bekman wrote:
>Any chance you could compare the logic of mod_status and Apache::VMonitor 
>and see why there is a difference?

Let me get my stick and turn off the lights.

>I'd suspect that what you see is some process IDs that used to be use by 
>Apache but now re-used by the OS, and the scoreboard gives the stale info.

Unlikely.  The errant processes all seem to be daemontools programs like 
supervise and svcscanboot.  Those processes are started long before apache 
starts, and they exist until after it goes away.

>I see that mod_status has all kind of  freshness checks. May be 
>Apache::VMonitor should do the same?

It couldn't hurt.

>I've just released Apache::Scoreboard 0.12 which exposes the vhost record 
>for mod_perl 1, and the newly released Apache::VMonitor 2.02 uses that 
>information.

VMonitor 2.02's `make test` is complaining loudly most of the time, failing 
tests 3-6.  It managed to succeed once, though.   If I go ahead and install 
it anyway, pulling up the VMonitor page causes the process to segfault.

Never a dull moment.