rl(4): autoselect/manual media == no carrier at boot time?
Hello, that's me again :) rl0: RealTek 8139 10/100BaseTX port 0xec00-0xecff mem 0xdf00-0xdfff irq 10 at device 5.0 on pci0 rl0: Realtek 8139B detected. Warning, this may be unstable in autoselect mode ^^ ^^ This is my detected NIC. It has always been detecting 100baseTX half-duplex, when I am actually plugged into a cool 100Mbps switch. So after a couple of tests, I decided that full-duplex was going fine and appended media 100baseTX mediaopt full-duplex to the ifconfig_rl0 line in my rc.conf, just in case. The card was actually working fine in this mode. Also given that rl(4) isn't so happy with autoselect mode if the proper PHY chip is not present (or something), I thought this would be a good thing. Now it happened that the server rebooted after a kernel panic (this is 5.0, I will be taking *this* subject to -current), but the problem here is that rl(4) detected no carrier : kernel: rl0: flags=8843UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST mtu 1500 kernel: inet 666.666.222.35 netmask 0xff00 broadcast 666.666.222.255 kernel: ether 00:50:fc:97:41:b4 kernel: media: Ethernet 100baseTX full-duplex kernel: status: no carrier My ISP says that they have not touched the UTP cable. After they rebooted the server again while trying to find why it was down, the same no carrier appears in the dmesg, so after a normal boot too. They claim that after this second reboot they unplugged and then plugged back the UTP connection and the NIC's led came to life and the network was up again (and still is, full-duplex). My question is, is it possible that setting the media to 100baseTX full-duplex could cause problems for carrier detection at boot time? Or is my ISP lying and they unplugged the network just right during the kernel panic and only plugged it back after a while? This doesn't sound likely... I do not feel like rebooting again right now just to see if using autoselect works, I'd rather not have to call them today to fix it once more ;-) Thanks in advance Olivier, always learning more thanks to you! To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
[Urgent] sendmail not echoing back during smtp connection
Help! I have a production server where sendmail suddenly stopped answering smtp connections! I have tried killing it and restarting it, but this doesn't change anything. The config didnt change, it just started to do this progressively, from what I can see in the logs with increasing numbers of [xxx.xxx.xxx.xxx] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA If I connect to port 25 the connection is established but nothing is echo'd, like no sendmail greeting, no reply whatsoever to smtp commands. So all MTAs are timing out connecting to my server! But the delivery still works fine, ie if I telnet to port 25 and give a valid mail from, rcpt to, data, etc, without waiting for SMTP answers, the mail is delivered fine in my inbox... How is this possible? I have tried everything, even when it doesn't make sense, like allowing all from all in IPFW, restarting sendmail many times, switching configs... short of rebooting the box of course, which probably wouldn't help anyway. LogLevel to 30 just yields a lot of intertwined : --- Mar 24 18:51:54 charon sm-mta[39105]: h2OHkian039105: Milter: no active filter Mar 24 18:51:54 charon sm-mta[39105]: h2OHkian039105: --- 220 charon.dom.tld ESMTP Sendmail 8.12.6/8.12.6; Mon, 24 Mar 2003 18:51:54 +0100 (CET) Mar 24 18:51:54 charon sm-mta[39105]: h2OHkian039105: --- 421 4.4.1 charon.dom.tld Lost input channel from tomts4.bellnexxia.net [209.226.175.10] Mar 24 18:51:54 charon sm-mta[39105]: h2OHkian039105: lost input channel from tomts4.bellnexxia.net [209.226.175.10] to MTA after startup Mar 24 18:51:54 charon sm-mta[39105]: h2OHkian039105: tomts4.bellnexxia.net [209.226.175.10] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA --- But this --- 220 charon.dom.tld ESMTP Sendmail 8... never actually reaches the client. (*but* from time to time there are successful connections in between!) Any ideas? I think I must be cursed or something... Olivier PS: please make sure you CC to me as my address subcribed to the list isnt getting the mails anymore To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
(SOLVED) Re: [Urgent] sendmail not echoing back during smtp connection
On Mon, Mar 24, 2003 at 06:58:21PM +0100, Olivier Dony wrote: Help! I have a production server where sendmail suddenly stopped answering smtp connections! I have tried killing it and restarting it, but this doesn't change anything. The config didnt change, it just started to do this progressively, from what I can see in the logs with increasing numbers of [xxx.xxx.xxx.xxx] did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA If I connect to port 25 the connection is established but nothing is echo'd, like no sendmail greeting, no reply whatsoever to smtp commands. So all MTAs are timing out connecting to my server! But the delivery still works fine, ie if I telnet to port 25 and give a valid mail from, rcpt to, data, etc, without waiting for SMTP answers, the mail is delivered fine in my inbox... Here is the explanation if someone ever gets the same problem: DNSBL's are somehow blocking every connection, so the solution is to remove every DNSBL from sendmail's config and restart it and the problem goes away. (A lot of thanks to the sendmail-questions mailing-list :-) I was using 4 of them : relays.ordb.org, list.dsbl.org, dun.dnsrbl.net and spam.dnsrbl.net. I will try and see which one(s) is(are) causing the problem when the mail traffic calms down a bit. This is probably due to the way DNSBL is implemented that nothing is logged regarding this problem and the client is just left there to timeout, and the timeout is really logged. This is what mislead me, but the progressive increase of the did not issue MAIL/EXPN/VRFY/ETRN during connection to MTA messages in the log should be a red flag for anybody using DNSBLs. Hope this can help someone. Thanks for reading ;) Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: pkgdb clone disks (second chance)
On Tue, Mar 18, 2003 at 12:18:16PM +0100, Erwane Breton wrote: No response first time. I try again. hi all First, i've installed perl5.6.1 with the system, and later i upgrade (install ?) to perl5.8.0. I ran the script use.perl port, and i deinstall perl5.6.1. And now always i want to upgrade a ports who need perl, i must run 'pkgdb -F' for fix the dependency with ... perl 5.6.1. when i install a ports with 'portinstall', i always see cleaning perl5.6.1. How i can tell to portinstall or make to use perl5.8.0 and remove perl5.6.1 ? That's what you did with 'use.perl port', and you can check that it appended a couple of lines about the perl version to use to /etc/make.conf. From now on make will use the port version when installing perl stuff, and perl scripts will too. Although ports that had been installed before might need to be rebuilt indeed, or have at least their dependencies fixed. # The second is a project For best results you should keep different questions on different e-mails with relevant subjects. I have a server running FreeBSD5.0 and client running win98 and i want to dump disk of client to create snapshot of the system for restore it later when it crashes :o). Like ghost in fact but on FreeBSD. thanks for idea too :) You can have a look at rsync from the ports, it's easy to use for snapshot backups. A good tutorial on using it for this very purpose is located at : http://www.mikerubel.org/computers/rsync_snapshots/ IIRC, it even explains how this can be used in conjunction with samba to backup windows stuff. Hope this helps. Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: Apache mod_ssl how to?
On Thu, Mar 20, 2003 at 12:44:01PM +0100, Andreas Widere Andersen wrote: Hi, I recently installed the apache-13-modssl port on my FreeBSD 4.8 RC system. I'm trying to set up a https virtual host without luck. I get this from my httpd-error.log: Ops, no RSA or DSA server certificate found?! I understand there's a missing certificate, but I can't seem to find a website or info about how to create one and do this with a full description of the process. Can someone in here please direct my to a place where I can find this info? A search on google will give hundreds of results about the subject. You can try: http://slacksite.com/apache/certificate.html And there's also https://ca.freeicp.org/webra/entry-level if you want someone to do the work for you ;) Hope this helps. Olivier PS: Please don't use reply on an existing thread to post a new question, as this will mess up the threads and diminish the visibility of your question too. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: Resource temporarily unavailable
On Tue, Mar 18, 2003 at 09:03:32AM +0100, Jan Blomqvist wrote: Hi! I have 800mhz 500mb node with freebsd 4.1, I often get this Resource temporarily unavailable I have problem to find out what kind of Resource its problem with, when I do top command everything looks ok lots of memory free and like 80% idle, and we have done some changes in the kernel options MAXMEM=(512*1024) options SHMMNI=64and also increased maxusers to 128.This is a example of error message: Hi, Just a couple of wild guesses : - I suppose you have no particular maxprocess limit or whatever for your login class in /etc/login.conf? You can always have a look at login.conf(5) - What does 'ps auxww' look like when this happens? Anything surprising there? 'top' only shows the first so much heavier processes, so you might not see everything. - On my box here I just ran a little test and with all limits set to 'unlimited' in login.conf I can trigger this error with a recursive shell script calling itself. It happens when about 475 processes have been spawned, and then they all die eventually and everything returns to normal. Hope this helps a bit, but people on the list have probably more information and also more relevant about this matter... good luck :-) Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Too many collisions on network?
Hello, I have a webserver serving a lot of requests (almost filling up 4 Mbits/s of http at peak time everyday). It's running Apache 1.3.27 on FreeBSD 4.7-RELEASE, and there are usually about 200-250 http processes running. Today I was looking at the various *stat tools to monitor it, and noticed a high number of network collisions ie : [EMAIL PROTECTED]:/root# netstat 5 input(Total) output packets errs bytespackets errs bytes colls 1598 0 190789 2035 02326238 659 1722 0 220532 2298 02958770 981 2547 0 347703 3078 02701052 1211 1937 0 274644 2461 03332772 1117 1619 0 235935 2169 02795248 725 1739 0 262005 2257 02076768 885 1461 0 255291 1781 02120986 537 1685 0 273239 1941 02144180 699 and so on... Earlier today there has been a time where the webserver was unreachable, with 1 out of 200 packets reaching it only, then everything went back to normal. Nothing unusual in the logs. Is there a way for me to see if the problem was the webserver itself or the network? And is this amount of collisions suspicious, given that the ISP is of course running a switched network? Thanks in advance for any ideas :-) Olivier PS: Lately I have been unable to post anything on the freebsd mailing-lists from my blacktrap.net e-mail address, the mails are accepted by the MX but seem to be discarded. I mailed [EMAIL PROTECTED] but did not receive any answer, maybe the mail was discarded too. What can I do about this? To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: Too many collisions on network?
On Sunday, 16 March, 2003 16:23, Simon Barner wrote: Are you running your network adaptor in full-duplex mode? Perhaps the device on the other end of the wire does only support half-duplex. Changing this increased the throughput from 20kb/s to almost 1mb/s on our internal network here (10baseT/UTP). I'm not really sure about this, here is the output of ifconfig related to the external interface : [EMAIL PROTECTED]:/root# ifconfig -m fxp0 fxp0: flags=8843UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST mtu 1500 inet xxx.xxx.191.100 netmask 0xfff0 broadcast xxx.xxx.191.111 inet xxx.xxx.191.101 netmask 0x broadcast xxx.xxx.191.101 ether 00:e0:81:23:c5:32 media: Ethernet autoselect (10baseT/UTP) status: active supported media: media autoselect media 100baseTX mediaopt full-duplex media 100baseTX media 10baseT/UTP mediaopt full-duplex media 10baseT/UTP media 100baseTX mediaopt hw-loopback It seems that fxp(4) supports full-duplex but that it is currently disabled, and I guess this is because the other end of the wire doesn't support it, since the media has been autoselected? Please correct me as I have never played a lot with ifconfig except for basic configutarion. I wouldn't be too eager to do tests with this setting since this is a production server and I have no direct physical access to it, so shutting down the only interface by mistake is *not* an option ;-) But if I was to change it, how would I go about this without shutting down fxp0? 'ifconfig fxp0 media 10baseT/UTP mediaopt full-duplex' ?? Thanks for your help :-) Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
mod_cgi not working at all/weird behaviour
Hello, I posted a problem some days ago about POST data not being passed to my cgi perl scripts. This only happens on my FreeBSD4.7-RELEASE server. Well in the process of troubleshooting I noticed that if I disabled mod_gzip (which is the only thing I don't have on my other working servers) my scripts stopped working completely with error 500. I tried then to reinstall everything. So I deinstalled perl, apache and what depends on it, deleted apache's conf file and reinstalled everything *but mod_gzip*. Well my cgi's aren't working a bit more than before! Anything I put in a cgi script ends with an internal error 500, and in the errorlog it always says : [Fri Mar 14 00:18:19 2003] [error] [client 212.68.212.45] Premature end of script headers: /home/oli/public_html/cgi-bin/test.cgi *Anything*, meaning that it fails too with a basic script like this : --- #!/bin/sh echo content-type: text/html echo htmlbodyhello/body/html --- (the empty line after content-type is there) My scripts are done by hand using vim, so no binary ftp xfer involved. All the scripts are chmod'd 0755, the containing directory too, and they give the expected output when run in the console. This is with apache 1.3.27_4 from the ports, mod_php4 4.3.1, and perl base or perl 5.6.1_11 from the ports (same result). I've tried doing tests to output this basic thing with sh, perl, php, etc.. everything gives the same 500 error when executed through CGI, while working ok in the console!! And guess what, if I reinstall mod_gzip, as soon as it is enabled and set to handle cgi-script with (mod_gzip_item_include mime ^text/html and mod_gzip_item_include handler ^cgi-script$) the scripts start to work, ie they give the expected output, but the first problem comes back : POST data is always empty. So I had to change my scripts to use GET for submitting forms, which is ugly and might even trigger query-string length problems I suppose. I'm completely lost, this doesn't seem to make sense at all, but is probably making sense if I knew where to look... The rest of the webserver is doing fine, basic html files and php are working ok, even with POST data! How could mod_gzip make my cgi scripts work, but only partly, and how come they aren't working at all without it, when working ok on my other FreeBSD webservers ?? Could someone help me or advice me before I completely lose my poor mind? ;-) I've tried to reinstall everything, change permissions etc.. Always the same result... Thanks a lot in advance if someone can provide any sort of help! Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: Too many collisions on network?
On Sunday, 16 March, 2003 21:40, Bill Moran wrote: Dan Nelson wrote: Input errors probably means that whatever you're plugged into does not support full-duplex (i.e. it's a hub), which is why autoselect selected half-duplex. You know, now that he brings this up, I tend to agree. It's possible that the hub you're plugging into is saturated: that will result in high collision rates and poor performance. You won't get anything better because that's the best the hub can do. If I'm right on this guess, you need to take it up with your ISP, as it's something they need to address. Ok I left this one in autoselect for now, and will contact my ISP tomorrow, and try and ask them to find solution for this high collision number problem... Thanks a lot everybody who answered :) Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Re: POST data not available in cgi perl scripts
On Wed, Mar 12, 2003 at 12:48:44PM +, Matthew Faircliff wrote:
Hello,
Hiya,
Have you tried using libapreq? I find that it is a superior method of transmitting
POST data. Find it in /usr/ports/www/p5-libapreq.
$my_posted_var = $r-parms{'my_var'};
Nope haven't tried it yet, because using CGI.pm was working for me on my other
servers, I never felt the need for a more clever thing, but I'll give it a try
for sure, this looks great.
Also, why not look into Mason (like emb_perl, only better). It handles POST data
like a charm.
I'll have a look at it right now too.
But anyhow I don't think those will solve the problem because the POST data
coming from the HTTP request doesn't even reach my script, although it has
been working all the time on my previous servers. So I guess the problem is
not in my method of getting/parsing the data, but in the process between the
client's request and the script getting it. And PHP scripts get the POST data
fine, argh :(
But thanks a lot for your suggestions, I'll try them as soon as I am done
compiling the perl port and re-installing all my perl modules.. Not that there
is a chance of it working better, since the system's perl works ok for POST
when trying it in the console. But I'm lost so trying anything I haven't
yet...
Thanks again Matt :)
--
Olivier
To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-questions in the body of the message
POST data not available in cgi perl scripts
I understand this might be a little off topic but I can only reproduce the
problem on one of my FreeBSD server, so it's not that much OT.
It's very simple, the data from a HTTP POST never reaches my CGI perl scripts.
There are no errors in the server log, the CGI executes fine, no debug with
CGI::Carp, etc.. only thing is that the POST data is always empty. And that
no matter what browser I use of course. Also I'm using CGI.pm, but handling
it by hand yields the same results. The POST DATA is always empty.
Though if I run my script with 'perl -c' in interactive mode and POST
something it works ok and the script gets my POST DATA without problem.
Displaying the %ENV shows the correct QUERY_STRING when using GET, but with
POST the POST DATA never makes it to the script.
This happens only on one of my FreeBSD : 4.7-RELEASE, apache 1.3.27, system
perl v5.005_03, whereas the same CGI scripts run fine on my other 5.0-RELEASE
and 4.4-RELEASE, with the same apache version, and perl different or the same.
I have no idea where or what I could look to solve the problem. Maybe others
have had the same issue or any perl/cgi gurus have any ideas for me?
I will be glad to provide any additional info or config files.
TIA a lot for any hints
--
Olivier
PS: An example of a very basic test that fails, I POST into this CGI script,
and the output of the POST data is empty, though CONTENT_LENGTH is correct :
FILE post.cgi
#!/usr/bin/perl -w
print Content-type:text/html\n\n;
read(STDIN, $buffer, $ENV{'CONTENT_LENGTH'});
print DATA: [$buffer]br\n; # - nothing output between the brackets here
@pairs = split(//, $buffer);
foreach $pair (@pairs) {
($name, $value) = split(/=/, $pair);
$value =~ tr/+/ /;
$value =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack(C, hex($1))/eg;
$FORM{$name} = $value;
}
print htmlbody;
foreach $key (keys(%FORM)) {
print $key = $FORM{$key}br;
}
# ^^^ here my output is always empty too
foreach $key (sort(keys %ENV)) {
print $key = $ENV{$key}br\n;
}
print /body/html;
/FILE post.cgi
To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-questions in the body of the message
Re: POST data not available in cgi perl scripts
On Wed, Mar 12, 2003 at 12:21:59AM +0100, Olivier Dony wrote: (...) Though if I run my script with 'perl -c' in interactive mode and POST ^^^ Sorry here I meant 'perl' in command line to get the CGI.pm interactive mode, and of course not 'perl -c', which I had used just before to check the syntax. something it works ok and the script gets my POST DATA without problem. (...) -- Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
Problem compiling 4.2-RELEASE kernel on another 4-4 computer
Hello, I've got some trouble compiling the 4.2-RELEASE kernel on my 4.4-RELASE box. The 4.2 one (dentaal) doesn't have enough hd space to do it. So I cvsup'd src-sys of tag=RELENG_4_0_2-RELEASE into /usr/DENTAAL on my 4.4 one. I tweaked GENERIC into DENTAAL, and ran config DENTAAL. I'm actually trying to add IPSEC support in the 4.2 kernel. Now trying to make depend : root@titan:/usr/DENTAAL/src/sys/i386/conf# ls ./ ../ DENTAAL GENERIC LINT NEWCARD root@titan:/usr/DENTAAL/src/sys/i386/conf# /usr/sbin/config DENTAAL Don't forget to do a ``make depend'' Kernel build directory is ../../compile/DENTAAL root@titan:/usr/DENTAAL/src/sys/i386/conf# cd ../../compile/DENTAAL root@titan:/usr/DENTAAL/src/sys/compile/DENTAAL# make depend cc -c -O -pipe -fomit-frame-pointer -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -fformat-extensions -ansi -nostdinc -I- -I. -I../.. -I/usr/include -D_KERNEL -include opt_global.h -elf -mpreferred-stack-boundary=2 ../../i386/i386/genas sym.c ../../i386/i386/genassym.c:76: warning: implicit declaration of function `__offsetof' ../../i386/i386/genassym.c:76: syntax error before `struct' ../../i386/i386/genassym.c:76: syntax error before `struct' ../../i386/i386/genassym.c:76: syntax error before `struct' ../../i386/i386/genassym.c:76: syntax error before `)' and a lot of other alike ... I assume there is a problem with this __offsetof() macro not being defined, but how do I fix the problem? Or is there anything I am doing wrong? Do I need to provide the kernel config file? Thanks a lot for any hints Olivier To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-questions in the body of the message
