Are these (this) stock kernels or suse-patched kernels? I don't wanna have to apply 50 patches just to install an rpm. On Friday 18 January 2002 15:09, James Bliss wrote:

You cannot use YOU to update the kernel because there are more steps necessary which YOU does not handle. You must download the RPM yourself, perform a rpm -Uhv <filename>.

After you perform the upgrade you then need to run mk_initrd and then lilo. Then you get the joy of rebooting (one of the few times you have to reboot in linux as opposed to havnig to reboot almost everytime you do anything in windows).

You should choose 2.4.16 since it is the most recent secure kernel for the 2.4 kernel versions. It includes some bug fixes. There had been an announcement for this some time back. I do not know if it was on this mailing list or on the security mailing list (I do not believe it was on this list).

Jim

01/18/02 03:03:15 PM, "Rafael E. Herrera" wrote:

...

I have noticed that there are RPM packages for the SuSE kernels in the update directories at the FTP site.

There are packages for version 2.4.10 and 2.4.16 for several SuSE distributions, e.g. 7.3, 7.2, etc.

There is no formal announcement nor any description on what these updates are for. YOU does not show it as an available update.

Are they security updates? Is there a reason I should choose 2.4.10 over 2.4.16? etc.

If someone has seen some description of these packages or announcement, I would appreciated it.

Rafael

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

This response is not only to this message, but to several messages after it as well. If you are installing the SuSE RPM from their ftp site then you are not going to need to install '50 patches' - you are thinking Microsoft, not linux. This post, as well as several others which I have seen recently, appear to be more attacks rather than requests for support. A request for support/help is usually more polite and usually avoids insults. These posts (sorry David, you are not as bad as several others, but you are a little too negative in your approach) need to realize that this list is a bunch of people who are willing to help where they can, but they will ignore you and not help you if you are being a JERK! Now, the most recent kernels which SuSE posts on their ftp site are stock kernels which will work with a SuSE configured machine (not in an environment which has been substantially 'hacked' (bad use of that term here)). To install one of the kernels you need to perform the steps which I and several other posters have mentioned. Please stop the PETTY insults, grow up, and realize that this is the world of open source. If you like Microsoft, then go pay them $60 per support request and talk with one of their high school level support representatives. If you like open source, then please be more considerate in your posts and ask your questions in a more polite manner. THIS IS GETTING RIDICULOUS AND DOWN RIGHT CHILDISH NOW!!! Sorry, venting a little, and tired of people who DEMAND help for free rather than realizing that I and many others on this list are GRACIOUSLY giving their time for free to help others (as well as asking for assistance themselves). Well, I am through ranting, have a good evening everyone. Jim 01/18/02 04:24:36 PM, David Grove wrote:

Are these (this) stock kernels or suse-patched kernels? I don't wanna have to apply 50 patches just to install an rpm.

On Friday 18 January 2002 15:09, James Bliss wrote:

...

You cannot use YOU to update the kernel because there are more steps necessary which YOU does not handle. You must download the RPM yourself, perform a rpm -Uhv <filename>.

After you perform the upgrade you then need to run mk_initrd and then lilo. Then you get the joy of rebooting (one of the few times you have to reboot in linux as opposed to havnig to reboot almost everytime you do anything in windows).

You should choose 2.4.16 since it is the most recent secure kernel for the 2.4 kernel versions. It includes some bug fixes. There had been an announcement for this some time back. I do not know if it was on this mailing list or on the security mailing list (I do not believe it was on this list).

Jim

01/18/02 03:03:15 PM, "Rafael E. Herrera" wrote:

...

I have noticed that there are RPM packages for the SuSE kernels in the update directories at the FTP site.

There are packages for version 2.4.10 and 2.4.16 for several SuSE distributions, e.g. 7.3, 7.2, etc.

There is no formal announcement nor any description on what these updates are for. YOU does not show it as an available update.

Are they security updates? Is there a reason I should choose 2.4.10 over 2.4.16? etc.

If someone has seen some description of these packages or announcement, I would appreciated it.

Rafael

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

I am surprised that you say that dhcpd client is broken in 2.4.16. I am on a cable modem and must run my external facing NIC using the dhcpd client and it runs flawlessly since I have upgraded to the 2.4.16 kernel. Can you enlighten me on what was broken so I can check to see if I am having any problems which I have not noticed. Thanks, Jim 01/18/02 03:50:31 PM, Ron Cordell wrote:

On Friday 18 January 2002 04:03 pm, Rafael E. Herrera wrote:

...

I have noticed that there are RPM packages for the SuSE kernels in the update directories at the FTP site.

There are packages for version 2.4.10 and 2.4.16 for several SuSE distributions, e.g. 7.3, 7.2, etc.

There is no formal announcement nor any description on what these updates are for. YOU does not show it as an available update.

Are they security updates? Is there a reason I should choose 2.4.10 over 2.4.16? etc.

If someone has seen some description of these packages or announcement, I would appreciated it.

Rafael

The kernels in the update FTP directory are the "official" kernel updates from SuSE for the distro that you are using. The kernels are supposed to be tested, but I am not certain what tests. For example, I don't know if Oracle is tested against kernel 2.4.16. This is opposed to the "unofficial" kernel updates in /people/mantel, which many people on this list use, but are not supported by SuSE. My main interest in getting a more recent kernel update, especially post 2.4.9, is the VM code in the kernel. I have seen some BIG improvements in this aspect of the kernel.

Now, I have noticed that some things are not completely correct in the 2.4.16 kernel. For example, dhcpd client is broken, and I have to use a fixed IP. This is supposedly fixed in the 2.4.17 kernel update, but I won't be updating from 2.4.16 until and if SuSE puts another update out for the 7.1 distro.

Ron Cordell

-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq and the archives at http://lists.suse.com

On Friday 18 January 2002 10:53 pm, Ron Cordell wrote:

On Friday 18 January 2002 05:06 pm, James Bliss wrote:

...

I am surprised that you say that dhcpd client is broken in 2.4.16. I am on a cable modem and must run my external facing NIC using the dhcpd client and it runs flawlessly since I have upgraded to the 2.4.16 kernel. Can you enlighten me on what was broken so I can check to see if I am having any problems which I have not noticed.

Thanks,

Jim

I could very well be full of &*(#... but when I updated to 2.4.16, I could no longer get an IP address using the DHCP client daemon. I did a search and found some discussion saying that the dhcpd client was broken in some cases in 2.4.16 and was fixed in 2.4.17. I can't find that thread just now, when I search again, but I haven't tried very hard...

-ronc

I just reset the network card to use DHCP using Yast2, at Yast2 reports a failure. However, the /var/log/messages shows the following: Jan 18 22:59:31 penguin kernel: bridge-eth0: lost peer eth0 Jan 18 22:59:31 penguin kernel: bridge-eth0: down Jan 18 22:59:42 penguin kernel: bridge-eth0: found peer eth0 Jan 18 22:59:42 penguin kernel: bridge-eth0: up Jan 18 22:59:45 penguin dhcpcd[9597]: broadcasting DHCP_REQUEST for 192.168.10.3 Jan 18 22:59:45 penguin dhcpcd[9597]: broadcastAddr option is missing in DHCP server response. Assuming 192.168.10.255 Jan 18 22:59:45 penguin dhcpcd[9597]: DHCP_ACK received from mars (192.168.10.1) Jan 18 22:59:45 penguin dhcpcd[9597]: infinite IP address lease time. Exiting Jan 18 22:59:49 penguin kernel: bridge-eth0: lost peer eth0 Jan 18 22:59:49 penguin kernel: bridge-eth0: down Jan 18 22:59:49 penguin kernel: bridge-eth0: found peer eth0 Jan 18 22:59:49 penguin kernel: bridge-eth0: up Jan 18 22:59:52 penguin dhcpcd[9639]: broadcasting DHCP_REQUEST for 192.168.10.3 Jan 18 22:59:52 penguin dhcpcd[9639]: broadcastAddr option is missing in DHCP server response. Assuming 192.168.10.255 Jan 18 22:59:52 penguin dhcpcd[9639]: DHCP_ACK received from mars (192.168.10.1) Jan 18 22:59:52 penguin dhcpcd[9639]: infinite IP address lease time. Exiting Jan 18 22:59:53 penguin startproc: startproc: exit status of /usr/sbin/rpc.lockd: 1 Jan 18 22:59:59 penguin kernel: eth0: no IPv6 routers present Jan 18 23:00:30 penguin startproc: startproc: exit status of /usr/sbin/rpc.lockd: 1 Jan 18 23:00:35 penguin PAM-unix2[8733]: session finished for user root, service su Jan 18 23:03:24 penguin su: (to root) rcordell on /dev/pts/1 Jan 18 23:03:24 penguin PAM-unix2[10687]: session started for user root, service su So, it looks like it may be working after all? -ronc

On Fri, 2002-01-18 at 23:13, Ron Cordell wrote:

On Friday 18 January 2002 10:53 pm, Ron Cordell wrote:

...

On Friday 18 January 2002 05:06 pm, James Bliss wrote:

...

I am surprised that you say that dhcpd client is broken in 2.4.16. I am on a cable modem and must run my external facing NIC using the dhcpd client and it runs flawlessly since I have upgraded to the 2.4.16 kernel. Can you enlighten me on what was broken so I can check to see if I am having any problems which I have not noticed.

Thanks,

Jim

I could very well be full of &*(#... but when I updated to 2.4.16, I could no longer get an IP address using the DHCP client daemon. I did a search and found some discussion saying that the dhcpd client was broken in some cases in 2.4.16 and was fixed in 2.4.17. I can't find that thread just now, when I search again, but I haven't tried very hard...

-ronc

I just restarted the machine with DHCP instead of a fixed address. The boot log shows:

Starting service dhcp client on eth0dhcpcd[260]: broadcasting DHCP_REQUEST for 192.168.10.3

dhcpcd[260]: broadcastAddr option is missing in DHCP server response. Assuming 192.168.10.255

dhcpcd[260]: DHCP_ACK received from mars (192.168.10.1)

dhcpcd[260]: infinite IP address lease time. Exiting

failed

However, I have an IP address, shown correctly by ipconfig. Why is DHCPD showing failed?

-ronc

I've been getting these messages on bootup with Kernels 2.4.0, 2.4.4, and 2.4.16 (all official SuSE SMP kernels), and it always tells me that the service has failed on startup....but I've never had any problem with networking. I'd suspected that it was my Netgear router (the DHCP server) since it hadn't caused any problems on the client side. - David A. Riggs

Hi Chris, Op vrijdag 18 januari 2002 23:39, schreef je:

...

 I don't understand why the mk_initrd and lilo sequence are not part of  the post installation scripts of the RPMs, since those steps must be  taken, anyway.

Because there's no way for a post-install script to know whether /etc/rc.config:INITRD_MODULES.  SInce the initrd isn't isn't

Is this the complete phrase you intended to send. It seems your interesting explanation about /etc/rc.config:INITRD_MODULES is not finished. Why wouldn't the rpm post script be able to process the /etc/rc.config file or to call mk_initrd?? -- Richard Bos For those without home the journey is endless

Chris, thanks for your explanation. Is it for the same reason that "services" (those part of /etc/rc.d) are not restarted when they are updated? Op zaterdag 19 januari 2002 00:03, schreef Christopher Mahmood:

* Richard Bos (allabos@freeler.nl) [020118 14:55]:

...

...

...

 I don't understand why the mk_initrd and lilo sequence are not part of the post installation scripts of the RPMs, since those steps must be taken, anyway.

Because there's no way for a post-install script to know whether /etc/rc.config:INITRD_MODULES.  SInce the initrd isn't isn't

Is this the complete phrase you intended to send. It seems your interesting explanation about /etc/rc.config:INITRD_MODULES is not finished.

Doh. "Because there's no way for a post-install script to know whether /etc/rc.config:INITRD_MODULES is correct." Thanks.

Ah ha. Does the user/administrator know ... ;)

...

Why wouldn't the rpm post script be able to process the /etc/rc.config file or to call mk_initrd??

Sure, mk_initrd already does source rc.config to get INITRD_MODULES. I mean, if you really want to do it all with one carriage return rpm -Uvh new_kernel.rpm && mk_initrd && lilo && init 6

It's not that I want it to be on 1 line, but when updating 50 packages it would be nice that mk_initrd/lilo are called when needed. Same will be valid for restarting services. This is especially true when doing a bulk update at many machines. It is of course possible to develop a wrapper around rpm, that does al this, isn't it. BTW: is /sbin/SuSEconfig --module executed by the rpm post script, or must this always be done afterwards by the user? -- Richard Bos For those without home the journey is endless

Op zaterdag 19 januari 2002 11:24, schreef Richard Bos:

It's not that I want it to be on 1 line, but when updating 50 packages it would be nice that mk_initrd/lilo are called when needed.  Same will be valid for restarting services.  This is especially true when doing a bulk update at many machines.  It is of course possible to develop a wrapper around rpm, that does al this, isn't it.

Would it be possible to create an rpm that contains info if processing is required after has been installed. Something like rpm -q --queryformat "%{TODOAFTER-RPM-INSTALLATION}\n", this way SuSE can provide the info and the user/sysadmin can build a wrapper around rpm to process the information SuSE provides. In case of k_dflt this would result in /sbin/mk_initrd and /sbin/lilo. -- Richard Bos For those without home the journey is endless

* Nadeem Hasan (nhasan@nadmm.com) [020118 15:01]:

I don't agree. You still need to run lilo because your kernel image (/boot/zImage) just changed, otherwise the system maynot boot next time around.

Of course, but the original question was "why doesn't the a post_install script run mk_initrd and lilo". I gave a reason why mk_initrd isn't run and without the new initrd there's no point in the post_install script running lilo. -- -ckm

* Nadeem Hasan (nhasan@nadmm.com) [020118 15:13]:

And my point was, you need to run lilo regardless of initrd. so, why not just run it anyway? I remember booting without running lilo after a kernel update. I didn't.

No, if you don't have the new initrd running lilo does nothing: step 1) install the kernel image, new modules, and System.map step 2) create the new initrd with the new modules and copy it into /boot step 3) create the new lilo map that knows about the new initrd and kernel You can't do step 3 before you've done 1 or 2. -- -ckm

* Nadeem Hasan (nhasan@nadmm.com) [020119 09:19]:

You are right, but you still miss my point. You are assuming that everyone uses and needs initrd. That is not true.

Since we have resierfs support complied as a module (not to mention scsi and raid controllers) everyone who uses reiserfs for their root fs, which I think is quite a few people, needs it. -- -ckm

Christopher Mahmood wrote:

* Rafael E. Herrera (raffo@neuronet.pitt.edu) [020118 14:20]:

...

I'm familiar with the installation process of these new kernels. What I'd like to see clearly posted in the ftp site is an announcement.

You didn't seem to have any trouble finding it. The descriptions of what is in each kernel package are given during the installation and in the manual.

I'm referring to a description of the kernel rpms posted as updates, like the INDEX file on any of the other directories (e.g. ftp://ftp.gwdg.de/pub/linux/suse/ftp.suse.com/suse/i386/update/7.3/a1/INDEX). The RPMs were put there for a reason, I'm only asking for those reasons to be shared with us.

...

The output of rpm -qp --changelog ... is useful, but who would guess it.

There's no need to guess it, it's documented in rpm(8).

yeah, yeah, RTFM :)

...

I don't understand why the mk_initrd and lilo sequence are not part of the post installation scripts of the RPMs, since those steps must be taken, anyway

Because there's no way for a post-install script to know whether /etc/rc.config:INITRD_MODULES. SInce the initrd isn't isn't automatically created there's no reason to run /sbin/lilo.

rc.config and lilo.conf could be parsed to find out if the machine is using the initrd mechanism and run the commands if necessary. lilo must be run in any case. The debian distro allows one to upgrade a kernel, although I'm not sure if it reloads the bootloader. Their scripts may be of help. I'm not completely sure, since I'm not very familiar with that distro. Raf