Subject:
|
Re: Load too high?
|
Newsgroups:
|
lugnet.admin.general
|
Date:
|
Fri, 15 Sep 2000 00:32:28 GMT
|
Viewed:
|
593 times
|
| |
|
|
In lugnet.admin.general, Todd Lehman writes:
> In lugnet.admin.general, Eric Joslin writes:
> > Just a couple of minutes ago (5:02pm Eastern by my system clock) I received
> > this error message:
> >
> > Server temporarily unavailable due to heavy load.
> > Please try again in a few minutes.
> >
> > several times while trying to load the page at URL:
> >
> > http://news.lugnet.com/?n=*,-20&v=c
> >
> > I saw some discussion between you and Larry Peniazek about it, but didn't
> > follow closely enough to get the details. Thought you might like to know
> > though.
>
> Thanks. My fault. I launched a process which forked a bunch of times to do
> some updates and things. I noticed it too, and raised the threshold. I set
> the threshold at a low conservative value initially. It's still a bit
> conservative now, but less so.
>
> --Todd
Umm, isn't that what MaxClients is for? The only thing I use load triggerd
shutoff for is the Virtual Explorer, since someone could (and has) pointed
a rude paralell robots.txt ignoring web-sucker at it and caused 40+ povray
renderings to start. In that case it only shuts off the Virtual Explorer.
Wouldn't MaxClients be exceeded long before loading the page in question
brought the server to a crawl?
Of course if something in the background did kick up the load very high
(like > 20) you might want an ohno page. I'd rather let the real pages
be served real slow than not at all. The assumption is that web hits aren't
going to make a significant difference in the load average wether it's 0.01 or
20.
KL
|
|
Message has 1 Reply:
 | | Re: Load too high?
|
| (...) No. Depending on what the processes are doing, you can have dozens of httpd processes all impacting the load average very negigibly, especially if they're blocked on I/O writes to the socket. --Todd (25 years ago, 15-Sep-00, to lugnet.admin.general)
|
Message is in Reply To:
| | Re: Load too high?
|
| (...) Thanks. My fault. I launched a process which forked a bunch of times to do some updates and things. I noticed it too, and raised the threshold. I set the threshold at a low conservative value initially. It's still a bit conservative now, but (...) (25 years ago, 14-Sep-00, to lugnet.admin.general)
|
4 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
