Mysterious hangs with Fedora
by rbp@netcanvas.com
Hello!
I'm experiencing a very weird situation with FC1.
I have this machine that was upgraded from RH9 to FC1, using yum. Everything
worked fine, no problem there.
When running RH9, this machine was rock solid and only rebooted for kernel
updates. But with FC1, the problems began.
The machine completely hangs, with no apparent reason whatsoever, at no specific
time. I tried lots of kernels.. my usual custom vanilla+grsec and RH9's 2.4.18
and Fedora's latest NPTL kernel. That appears to make no difference. However,
there was this occasion when i noticed the machine hanged two days in a row at
exactly the same time, when running cron.daily (updatedb?!). But i never got to
reproduce that sistematically. Yesterday i stressed tested it with a full
backup (tar+gzip) and kernel compilations at the same time. Went great. Then i
went to bed, to find in the morning it crashed around 30mins after that test,
when idle.. I can have a week uptime, or 3 hangs in a day, depends on it's
mood. The machine does pretty much the same thing all the time: mail and web,
nothing fancy and nothing else.
The hardware is a Tyan 1U rackmount, 2.4 P4, 1GB DDR, two 120GB IDE disks in
RAID1 (Promise FastTrack controller onboard). Hardware worked before for
months, no problem. All fans ok, so overheating does not seem like a
possibility. The only changed variable is the upgrade from RH9 to FC1. However,
same behaviour with very different kernels, that also confuses me. Some quirk
with glibc maybe?
Anyone here experiencing something similar? Any thoughts? Before i switch back
to RH9 and see if it helps..
Thanks in advance
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
20 years, 1 month
core 2 test 2 : fonts
by Raghu Angadi
Hi,
I just upgraded from test1 to test2.
One thing I noticed is that font rendering seems to have changed in xterm,
xemacs (apps that use traditional fonts). I run xterm with 'courier new' and
the font looks awful after the upgrade. No config differences that I know of.
Do you know what's going on?
xterm command:
xterm +sb -fn -*-courier new-medium-r-*-*-*-140-*-*-*-*-iso8859-*
thanks,
Raghu.
__________________________________
Do you Yahoo!?
Yahoo! Finance Tax Center - File online. File on time.
http://taxes.yahoo.com/filing.html
20 years, 1 month
Copy with SCP - low performance
by Horacio Reyes
Hi,
I am using scp to copy large amount of small files between 2 servers
with a firewall in the middle, but the copy takes all CPU, this is the
situation:
- The files are copied in batch of 100 tif images, in a directory with
more than 5 million files.
- the server that initiate the scp is a 1.2Ghz P3 with 2G in ram, the
receiver is a double Xeon 2.6ghz with 2G in ram. The first server takes
no time to send the file, but the receiver takes more than 200segs to
copy. They are in a 100mb lan.
- the top comand shows 98% free cpu before the scp, when scp start it
takes 100% of the 2 xeon cpu. The problem is not the ram, both servers
have more than 1G free when scp starts.
I tried with all the SCP flags (-q, -c blowfish, etc), and nothing
works....
Regards,
Horacio Reyes Rios
Eng. Director.
**************************************************************************
Nota: La informacion transmitida a traves de este correo electronico es confidencial y esta dirigida unicamente a su destinatario. Su reproduccion, lectura o uso esta prohibido a cualquier persona o entidad diferente. Aunque Computec S.A. ha realizado su mejor esfuerzo para asegurar que el presente mensaje y sus archivos anexos se encuentran libre de virus y defectos que puedan llegar a afectar los computadores o sistemas que lo reciban, no se hace responsable por la eventual transmision de virus o programas daninos por este conducto, y por lo tanto es responsabilidad del destinatario confirmar la existencia de este tipo de elementos al momento de recibirlo y abrirlo. Ni Computec S.A. ni ninguna de sus divisiones o dependencias aceptan responsabilidad alguna por eventuales danos o alteraciones derivados de la recepcion o uso del presente mensaje.
**************************************************************************
Note: The information transmitted through this E-mail is confidential and is intended to reach only its addressee. Reproduction and usage of this message or its contents, in whole or in part, are strictly forbidden without the prior written consent of Computec S.A. Computec S.A. has made its best efforts to ensure that this message and any attached files are free of any virus or other potentially harmful content, but makes no representations to this effect. Addressee shall assume full responsibility for ensuring that opening or reading this message and any attachments will not result in harm to their system. Neither Computec nor any of its divisions accepts any responsibility for damages arising from opening or reading this message or any attachments thereto.
**************************************************************************
20 years, 1 month
snort
by Rajesh Ganesan
Hi,
I have recently switched from Suse. In suse, by changing /etc/sysconfig/snort,
we can activate snort dynamically when dialup connection is established.
These are the values.
SNORT_INTERFACE=ppp0
SNORT_ACTIVATE=yes
SNORT_AUTO=yes
In snort of fedora, these values are not present and even if we add these,
snort does work. Any work around? Or am I missing something?
Regards,
Rajesh
20 years, 1 month
Re:Modem Lights
by Luc Lessard
Make sure that you have /sbin/ifup and /sbin/ifdown scripts available
(exec. permission and all.). Then in the Modem Light configuration do:
/sbin/ifup "Network device name or alias" and /sbin/ifdown "Network
device name or alias". In my case it is /sbin/ifup Dialup_ISP and
/sbin/ifup Dialup_ISP. "Dialup_ISP" is a nickname that I have selected
for the modem network connection in the "Redhat-network-configuration" tool.
This should enable the start an stop of your modem network.
My lock file was LOCK..ttyS0. This enable the connection light and timer.
And finally make sure that you device is "ppp0". This gives you the
traffic lights and graph.
Hope this made any sense.
Good luck.
Luc
Dean Mumby wrote:
check that it is pointing to the correct lock file
Regards
Dean
Dave Smith wrote:
> Has anyone been able to get the Modem Lights application working
> properly? I've got as far as getting it to flash but not the vital
> part of conneting and disconnecting.
>
> Thanks.
>
> Dave
20 years, 1 month
Managing services
by Fabrício Santos
Viva a todos,
Since at install time of my FC1 I did a pretty dumb instalation, Apache and
Samba were not automatically configured to start at boot up and I have done
it manually after. What I did was going to /etc/rc3.d and issued the
following command:
ln -s /etc/init.d/httpd S35httpd
Now I have a couple of questions and a couple of assumptions that I believe
true but am not sure about. So answers to my questions or links to further
reading will be most welcome.
1. service command.
I know a service in /etc/init.d (or /etc/rc.d/init.d/) may be started using
the service command but I couldn't find any proper docs on it. No man page
available. No --help switch available. Only the following is displayed when
the command is run:
Usage: service < option > | --status-all | [ service_name [ command
| --full-restart ] ]
Are there any more extensive docs to this utility?
2. /etc/rcX.d
Inside the rcX.d directories, there are files named K99servicename and
S99servicename. I supose the K stands for Kill at reboot or halt time, and S
stands for Start at bootup time. Then the numbers are just a way of sorting
the way the services are started or stopped. Is this correct? Where can I
learn more on these matters?
3. ln -s /etc/init.d/httpd S35httpd
Is there any FC1 text based utility that may create these links
automatically instead of forcing the way manually like I did?
Thanks for your help!
-fs
20 years, 1 month
Re: Just updated to Fedora, won't boot in SMP
by Klaus Steinberger
> I just upgraded to Fedora 1 from RH9. Everything is working well except
> when I try to boot into the SMP kernel. Right before running the INIT
> process the server just hangs. When I chose the non-smp option, it blows
> right by the same spot, and give the standard "Setting INIT level 3"
I have the same problem with an Fuijtsu-Siemens RX300 Server. I did a fresh
install on it, and booting up with SMP Kernel did hang somewhere were it mounts
the local disks. I'm not sure as I didn't dig too deep into it, but I suspect
the aacraid driver. Do you also have an Adaptec Raid controller with the aacraid
driver?
Maybe you could run you're system with the latest rh9 smp kernel until the
problem get's fixed.
I did try the Kernel from Fedora Core 2 Test1 as the machine is currently not in
production. The kernel from fc2t1 runs well also in SMP mode, but has a problem
with autosensing on broadcom tg3 driver, but the update kernel
2.6.3-2.1.253.2.1smp runs very well. Of course I would not suggest that for an
production system.
Sincerly,
Klaus Steinberger
--
Klaus Steinberger Maier-Leibnitz Labor
Phone: (+49 89)289 14287 Am Coulombwall 6, D-85748 Garching, Germany
FAX: (+49 89)289 14280 EMail: Klaus.Steinberger(a)Physik.Uni-Muenchen.DE
URL: http://www.physik.uni-muenchen.de/~k2/
In a world without Walls and Fences, who needs Windows and Gates
20 years, 1 month
Help needed tracking out hardware problem...
by Mauri Sahlberg
Hi,
As the guarantee of my laptop nears it's end the laptop has started to
have unwanted symptoms. Namely it shuts itself down without no apparent
reason. My guess is that it suffers from overheating and bails out.
Usually I get this after an hour of work and running a long sequence of
database queries (selects and inserts, about 20000 inserts and selects).
I've been told that the board and processor probably have built in
sensors for heat and several other goodies that could help me to
diagnose the problem but I have no clue whatsoever how to read them. I
guess I need something compiled in kernel and a suitable client program
to record and display the values in real time. Any suggestions or
keywords to throw in for google?
Regards,
--
Mauri "mos" Sahlberg Pretax Systems Oy +358 207 44 2228
Technology Evangelist Pääskylänrinne 8 +358 207 44 2201
Bsc Computer Science FIN-00500 Helsinki www.pretax.net
Development Manager Finland
20 years, 1 month
Re: Fedora core 1 sendmail problems
by Homer Sapions
Current status - still not working. I did not manage to work on it much last
night but I intend to try again tonight. Fortunately this is a personal
web/mail server so inbound mail is for myself and a few other family members
only.
Port 25 is not being blocked at the ISP or at the linksys. I can send mail
out with no problems, but can not receive mail. Traffc _is_ definitely
getting to the server on port 25. When I turn on tcpdump or ethereal
watching port 25 I see inbound connections, and I see my server attempting
to respond. Alexander Dalloz spent a lot of time attempting to help me debug
this, doing as I had done previously from another ISP, telnetting to my
server on port 25. He did not get an immediate disconnect, but it took a
number of seconds. During this time, I could see via tcpdump him connecting
to me, and my server attempting to respond, but he never got the ACK back
from my server.
iptables is not running on the server, and sendmail is not behind
tcp-wrappers. I am not aware of any routing problems, because outbound mail
works fine, and all of my web pages are working fine, including
Squirrelmail, CGI scripts etc. I have nothing in /etc/hosts.allow or
/etc/hosts.deny, and I modified my sendmail.mc to comment out the DaemonPort
options listing only the loopback IP address, then built a new sendmail.cf
with make -C /etc/mail, and restarted the sendmail daemons.
Last night I uninstalled iptables and tcp-wrappers to be totally sure there
is no correlation, but did not get as far as rebooting and testing some
more. I had other problems with my cable modem connection and loss of all
internet access for a while. I will try rebooting and testing further as
soon as I get home tonight. I will take a look at your how-to doc also and
see if I can pick up any possible problems from that.
>From: "Rodolfo J. Paiz" <rpaiz(a)simpaticus.com>
>Reply-To: For users of Fedora Core releases <fedora-list(a)redhat.com>
>To: For users of Fedora Core releases <fedora-list(a)redhat.com>
>Subject: Re: Fedora core 1 sendmail problems
>Date: Tue, 30 Mar 2004 08:58:46 -0600
>
>At 13:26 3/29/2004, you wrote:
>>I'm running a server - not a client. On the server I need to be able to
>>both send and receive mail using SMTP on port 25 to/from other servers.
>>I'm not catering for POP mail on port 110.
>
>Homer,
>
>I've lost track of what's going on here. What is your current status and
>problem? Are you sure it's not a network issue (firewall, router, ISP
>blocking the port, etc.)? Try also taking a look at the "Sendmail SMTP AUTH
>HOWTO" on my website [1]. While the concept is not exactly what you want,
>the steps shown for enabling receive from the network should work for you.
>
>[1] http://www.simpaticus.com/linux
>
>Only note that the sendmail.mc included there shows this line:
> * define(`confAUTH_OPTIONS', `A')dnl
>In reality, the Fedora installation of Sendmail has `A p' instead. Remember
>to remove the "p" or else it won't work; this is an error which I have
>regrettably not fixed yet in the document.
>
>Cheers,
>
>
>--
>Rodolfo J. Paiz
>rpaiz(a)simpaticus.com
>http://www.simpaticus.com
>
>
>--
>fedora-list mailing list
>fedora-list(a)redhat.com
>To unsubscribe: http://www.redhat.com/mailman/listinfo/fedora-list
_________________________________________________________________
Find a broadband plan that fits. Great local deals on high-speed Internet
access.
https://broadband.msn.com/?pgmarket=en-us/go/onm00200360ave/direct/01/
20 years, 1 month