<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"> <html> <head> <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type"> </head> <body bgcolor="#ffffff" text="#000000"> Eli, try to use nagios-2.0b4, it doest give any errors to me so far. Nagios 2.3 seems to be generating more errors in the CGIs than the previius 2.2. I will try to find a pattern on this and look in the code for memory leeks and th like. []s. Eli Stair wrote: <blockquote cite="midC08690FD.279FC%25estair@ilm.com" type="cite"> <title>Re: [Nagios-devel] Re: Nagios 2.3 internal server error.</title> Sorry for the clutter, my earlier post was too optimistic... Total of 5000 requests via elinks for a host detail shows only 3 500 (internal server error) issues on the client side. At the same time, this generated 63 “Premature end of script headers: status.cgi” errors in the apache logs. Only one revealed the referrer URL: [Tue May 09 16:23:57 2006] [error] [client 10.73.16.108] Premature end of script headers: status.cgi, referer: <a class="moz-txt-link-freetext" href="https://monitor02/nagios/cgi-bin/status.cgi?hostgroup=deathstar-opteron-850-32G&style=overview">https://monitor02/nagios/cgi-bin/status.cgi?hostgroup=deathstar-opteron-850-32G&style=overview</a> During this period of testing (several hours, 5000+ checks with links, and 10 windows open with various UI views refreshing every 5 minutes), only this one verbose message, the rest to the effect of: [Tue May 09 16:23:52 2006] [error] [client 10.73.16.108] Premature end of script headers: status.cgi And through the entirety firefox received four 500 pages, links three, and yet 63 “premature” errors were generated by the CGI’s. Still looks broken, my bad for being excited. Will compiling with any of the debug flags set cause the CGI’s to output more useful info, or are they only for the nagios daemon as it seems to be? /eli On 5/9/06 2:31 PM, "Eli Stair" <a class="moz-txt-link-rfc2396E" href="mailto:estair@ilm.com"><estair@ilm.com></a> wrote: <blockquote>Hmm, missed this. Gave this code a shot and am still seeing the problem, though it seems at a _MUCH_ lower rate of frequency. Of 1200 hits I got only 3 500’s returned by the client (previous failure rate was around 1:40). The other significant change is I’m not seeing segfaults reported by the CGI’s, nor the “premature end of script headers” message in the apache logs that used to correspond with these 500’s. I’m guessing this fixed the major problem and the symptoms of it (segv’s, script header issue)... Unless something else was changed, I’d say it’s “mostly fixed”, or at least better, likely due to the content_length issue? Thanks devs, I had actually given up hope that this would be tracked down and addressed. This is great news (pre-emptively). No idea if this was randomly spotted or someone went looking for it due to my (and others’) reports, but either way I appreciate it. Cheers, /eli 2.3 - 05/03/2006 * Bug fix for negative HTTP content_length header in CGIs On 5/9/06 5:22 AM, "Alessandro Ren" <a class="moz-txt-link-rfc2396E" href="mailto:alessandro.ren@opservices.com.br"><alessandro.ren@opservices.com.br></a> wrote: <blockquote> I've updated to nagios 2.3 and I am still getting the internal server error from time to time in the CGIs refresh. Eli, have you tried the 2.3 already? Just to let the list know. </blockquote> </blockquote> </blockquote> </body> </html>