Re: os-linux/3343: Server dies after 1-20 hours of usage.
The following reply was made to PR os-linux/3343; it has been noted by GNATS. From: Mark Herman II [EMAIL PROTECTED] To: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Subject: Re: os-linux/3343: Server dies after 1-20 hours of usage. Date: Sun, 21 Feb 1999 18:40:47 -0600 [EMAIL PROTECTED] wrote: [In order for any reply to be added to the PR database, ] [you need to include [EMAIL PROTECTED] in the Cc line ] [and leave the subject line UNCHANGED. This is not done] [automatically because of the potential for mail loops. ] [If you do not include this Cc, your reply may be ig- ] [nored unless you are responding to an explicit request ] [from a developer. ] [Reply only with text; DO NOT SEND ATTACHMENTS! ] Synopsis: Server dies after 1-20 hours of usage. Comment-Added-By: lars Comment-Added-When: Thu Feb 11 12:51:43 PST 1999 Comment-Added: [This is a standard response.] This Apache problem report has not been updated recently. Please reply to this message if you have any additional information about this issue, or if you have answers to any questions that have been posed to you. If there are no outstanding questions, please consider this a request to try to reproduce the problem with the latest software release, if one has been made since last contact. If we don't hear from you, this report will be closed. If you have information to add, BE SURE to reply to this message and include the [EMAIL PROTECTED] address so it will be attached to the problem report! Hi, The server still dies, but we did find something that may be contributing to it. We just are not sure why. He runs a mailbag script that mails the traffic of his bulletin board to its subscribers every hour. If he disables this script, the web server doesn't die. We haven't found anything in this script that we believe would kill the web server, but if you would like to see it, I can forward it to you. Mark
Re: os-linux/3343: Server dies after 1-20 hours of usage.
The following reply was made to PR os-linux/3343; it has been noted by GNATS. From: Mark Herman II [EMAIL PROTECTED] To: [EMAIL PROTECTED] Cc: [EMAIL PROTECTED] Subject: Re: os-linux/3343: Server dies after 1-20 hours of usage. Date: Thu, 05 Nov 1998 00:14:10 -0600 Hi, Thanks for the quick reply. Here are the settings you asked for: MaxRequestsPerChild is set to 64. It was set to 30, but I increased it. This seemed to make the server take longer to die. MaxClients is set to 150 MinSpareServers is 5 MaxSpareServers is 10 BTW, we are running a custom transfer log, but we weren't when the problem started. I noticed another message in the database regarding custom logs, but this shouldn't be the cause of the problem. There were no unusual messages in the error log file. I also checked the syslog messages file, and I did notice several messages about possible SYN floods, but the times don't appear to correspond with the unresponsiveness of httpd. Thanks [EMAIL PROTECTED] wrote: [In order for any reply to be added to the PR database, ] [you need to include [EMAIL PROTECTED] in the Cc line ] [and leave the subject line UNCHANGED. This is not done] [automatically because of the potential for mail loops. ] [If you do not include this Cc, your reply may be ig- ] [nored unless you are responding to an explicit request ] [from a developer. ] [Reply only with text; DO NOT SEND ATTACHMENTS! ] Synopsis: Server dies after 1-20 hours of usage. State-Changed-From-To: open-feedback State-Changed-By: lars State-Changed-When: Wed Nov 4 18:30:46 PST 1998 State-Changed-Why: Are there any messages in your error log? What are your MaxClients, MaxRequestsPerChild and Min/MaxSpareServers settings? Release-Changed-From-To: 1.2.6 and 1.3.3-1.3.3 Release-Changed-By: lars Release-Changed-When: Wed Nov 4 18:30:46 PST 1998
os-linux/3343: Server dies after 1-20 hours of usage.
Number: 3343 Category: os-linux Synopsis: Server dies after 1-20 hours of usage. Confidential: no Severity: critical Priority: medium Responsible:apache State: open Class: sw-bug Submitter-Id: apache Arrival-Date: Wed Nov 4 14:40:00 PST 1998 Last-Modified: Originator: [EMAIL PROTECTED] Organization: apache Release:1.2.6 and 1.3.3 Environment: Red Hat Linux with kernel version 2.0.35. It has been patched to glibc2.0.7. I am using gcc2.7.2.3, but the 1.2.6 server we were using came with Red Hat Linux. Description: There is no core dump. The server-status report from the 1.2.6 apache server would show processes marked as running although they no longer existed. Once it reached the maximum number of requests for a child, it would kill the child, but it would still appear as running in the server-status report. It would then start a new process, since it can't use the dead ones. After several hours it would reach the maximum number of processes and stop answering requests. The 1.3.3 version does not exibit the same behavior in the server status report, but it still dies. How-To-Repeat: I don't know what is causing it, so I can't repeat it. You can access the server-status report at the following URL: http://www.v6fbody.com/server-status/ The username and password are: username: apache password: group The website address is http://www.v6fbody.com/ Fix: We started a cron job to restart the server on a periodic basis. Audit-Trail: Unformatted: [In order for any reply to be added to the PR database, ] [you need to include [EMAIL PROTECTED] in the Cc line ] [and leave the subject line UNCHANGED. This is not done] [automatically because of the potential for mail loops. ] [If you do not include this Cc, your reply may be ig- ] [nored unless you are responding to an explicit request ] [from a developer. ] [Reply only with text; DO NOT SEND ATTACHMENTS! ]
protocol/875: force-response-1.0 bug
Number: 875 Category: protocol Synopsis: force-response-1.0 bug Confidential: no Severity: serious Priority: medium Responsible:apache (Apache HTTP Project) State: open Class: sw-bug Submitter-Id: apache Arrival-Date: Wed Jul 16 12:50:01 1997 Originator: [EMAIL PROTECTED] Organization: apache Release:1.2.X Environment: BSDI 3.0. I believe it will occur on any OS. I also experience it on Linux. Description: I run the JCount Java access counter at http://www.jcount.com/. We included the BrowserMatch lines that were suggested in the FAQ. This counter works properly on all browsers I've tried except Internet Explorer 4.0 Preview Release 2. The BrowserMatch lines tell the server to send back HTTP/1.0 response headers to IE 4.0 PR 2. While searching for the cause of this error, I tried telnetting directly into the server and typing the http requests myself. I've found that using the force-response-1.0 directive sends back the HTTP/1.0 header, but HTTP/1.1 encoded information. The specific problem that this is causing me is that if I call my CGI scripts in IE 4.0 PR 2, it has an extra number before the beginning of my program's output. It would seem that any Java applet that uses CGI to communicate with a server runs a chance of running into this problem on this browser. I think I will find a simple fix in my particular case, but I still think this should be addressed. How-To-Repeat: telnet into www.jcount.com port 80, and type the following: GET /cgi-bin/counter2.cgi?secondary_exposure=trueincrement=falsecounter_id=30 HTTP/1.1 Host: www.jcount.com I have set the force-response-1.0 environment variable to help find the problem. I will be setting this back to normal soon, but for now, the output will be: HTTP/1.0 200 OK Then what appears to be a HTTP/1.1 header will follow. After that, there will be a blank line, then some hexadecimal number, which is the cause of my problems. In my case, my program parses the output based on which line it is on. I am assuming that this new number has something to do with the HTTP/1.1 response, because it doesn't show up if I request the information using HTTP/1.0. Fix: Instead of having the force-response-1.0 directive just change the first line of the header sent back, send back a true HTTP/1.0 response Audit-Trail: Unformatted: