I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
Also, I'm using a Thinkpad T42 (2378RAU) and the wireless card shows up as a weird device after boot:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
- Julian
Julian C. Dunn schrieb:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
same problem here https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=185528
Also, I'm using a Thinkpad T42 (2378RAU) and the wireless card shows up as a weird device after boot:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
does this also happens with older kernels? seems like a kernel issue.
- Julian
dragoran wrote:
Julian C. Dunn schrieb:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
same problem here https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=185528
I added this as blocker as it may break upgrades from older versions.
From: "DI Mario Bruckschwaiger" mario@targetdevelopment.at
dragoran wrote:
does this also happens with older kernels? seems like a kernel issue.
It does not happen with kernel-2.6.15-1.1826.2.9_FC5 (everything else from rawhide) on my x86_64 laptop.
Not happened with :
Dependencies Resolved
============================================================================ = Package Arch Version Repository Size ============================================================================ = Installing: kernel i686 2.6.15-1.2054_FC5 development 13 M Updating: ORBit2 i386 2.14.0-1 development 247 k initscripts i386 8.31.1-1 development 1.2 M kudzu i386 1.2.34.3-1 development 401 k libbonobo i386 2.14.0-1 development 475 k python-urlgrabber noarch 2.9.8-2 development 119 k Removing: kernel i686 2.6.15-1.1955_FC5 installed 33 M
inittab still inittab and not found an rpmsaved one.
Danny.
Danny Terweij - Net Tuning | Net wrote:
From: "DI Mario Bruckschwaiger" mario@targetdevelopment.at
dragoran wrote:
does this also happens with older kernels? seems like a kernel issue.
It does not happen with kernel-2.6.15-1.1826.2.9_FC5 (everything else from rawhide) on my x86_64 laptop.
Not happened with :
Dependencies Resolved
============================================================================ = Package Arch Version Repository Size ============================================================================ = Installing: kernel i686 2.6.15-1.2054_FC5 development 13 M Updating: ORBit2 i386 2.14.0-1 development 247 k initscripts i386 8.31.1-1 development 1.2 M kudzu i386 1.2.34.3-1 development 401 k libbonobo i386 2.14.0-1 development 475 k python-urlgrabber noarch 2.9.8-2 development 119 k Removing: kernel i686 2.6.15-1.1955_FC5 installed 33 M
inittab still inittab and not found an rpmsaved one.
Danny.
have you ever edited inittab by hand?
From: "dragoran" dragoran@feuerpokemon.de
It does not happen with kernel-2.6.15-1.1826.2.9_FC5 (everything else from rawhide) on my x86_64 laptop.
Not happened with :
[cut]
Dependencies Resolved
have you ever edited inittab by hand?
No, installed FC5T3 less than 48 hours ago.
Downloaded the disc1, so i though should be enough for a "base" CLI only install. after disk1, it asked disc2 for a few packages, disc3 for a few packages, dics4 for 1 package and disk5 for 1 package (would be nice a "base" install with only disc1 needed). Afterthat uninstalled a lot unwanted packages. It ended up in 400mb disk usage. Still tooo much but dont know what to delete more (just want a kernel , and mysql 5 running for playing and testing for migrate mysql 3.x based apps. after that i did yum update. got a new kernel. rebooted it and today ran yum update again.
Enough information i guess :)
Danny.
On 03/15/2006 Danny Terweij - Net Tuning | Net wrote:
Afterthat uninstalled a lot unwanted packages. It ended up in 400mb disk usage. Still tooo much but dont know what to delete more (just want a kernel , and mysql 5 running for playing and testing for migrate mysql 3.x based apps. after that i did yum update. got a new kernel. rebooted it and today ran yum update again.
In the package selection screen of the installer, deselect everything except for the Mysql Database server (or whatever we call it). You'll get the packages of that group and their deps. Not much else (like possibly not even yum)
On Wed, 2006-03-15 at 16:28 +0100, dragoran wrote:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
does this also happens with older kernels? seems like a kernel issue.
Yes, it did happen with older fc5t3 kernels although I never mentioned it here before.
- Julian
Julian C. Dunn (julian.dunn@devlin.ca) said:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
I cannot reproduce this here. Moreover, I cannot see *how* this could happen.
/etc/inittab is marked %config(noreplace) in both the old and new packages.
In the rpm handling code, this should *never* end up with: old file -> rpmsave, no new file.
Also, I'm using a Thinkpad T42 (2378RAU) and the wireless card shows up as a weird device after boot:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
You have a wired ethernet that is configured to be eth0, and your wireless module was loaded first, so it was moved out of the way.
Bill
On Wed, 2006-03-15 at 13:38 -0500, Bill Nottingham wrote:
Julian C. Dunn (julian.dunn@devlin.ca) said:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
I cannot reproduce this here. Moreover, I cannot see *how* this could happen.
/etc/inittab is marked %config(noreplace) in both the old and new packages.
In the rpm handling code, this should *never* end up with: old file -> rpmsave, no new file.
Hmm. I looked into this further and it looks like somehow the yum update *removed* the old initscripts, which would explain the behaviour... see the excerpt from my /var/log/yum.log:
Mar 15 09:01:49 Erased: initscripts Mar 15 09:01:53 Updated: valgrind.i386 1:3.1.0-2 Mar 15 09:02:23 Installed: kernel-devel.i686 2.6.15-1.2054_FC5 Mar 15 09:02:24 Updated: grub.i386 0.97-5 Mar 15 09:02:26 Updated: initscripts.i386 8.31.1-1
I have no idea why it would do this.
Also, I'm using a Thinkpad T42 (2378RAU) and the wireless card shows up as a weird device after boot:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
You have a wired ethernet that is configured to be eth0, and your wireless module was loaded first, so it was moved out of the way.
Okay - is that expected behaviour? (I'm currently running without wireless configured, but I'm not suppressing the load of the kernel module or anything like that)
- Julian
Julian C. Dunn (julian.dunn@devlin.ca) said:
I cannot reproduce this here. Moreover, I cannot see *how* this could happen.
/etc/inittab is marked %config(noreplace) in both the old and new packages.
In the rpm handling code, this should *never* end up with: old file -> rpmsave, no new file.
Hmm. I looked into this further and it looks like somehow the yum update *removed* the old initscripts, which would explain the behaviour... see the excerpt from my /var/log/yum.log:
Mar 15 09:01:49 Erased: initscripts Mar 15 09:01:53 Updated: valgrind.i386 1:3.1.0-2 Mar 15 09:02:23 Installed: kernel-devel.i686 2.6.15-1.2054_FC5 Mar 15 09:02:24 Updated: grub.i386 0.97-5 Mar 15 09:02:26 Updated: initscripts.i386 8.31.1-1
I have no idea why it would do this.
Neither do I. Although... weird. Can you post/attach to bugzilla your whole log?
Also, I'm using a Thinkpad T42 (2378RAU) and the wireless card shows up as a weird device after boot:
dev8569 Link encap:Ethernet HWaddr 00:0E:35:FC:39:23
Does anyone know why?
You have a wired ethernet that is configured to be eth0, and your wireless module was loaded first, so it was moved out of the way.
Okay - is that expected behaviour? (I'm currently running without wireless configured, but I'm not suppressing the load of the kernel module or anything like that)
Yes. If you configre the wireless, it will be changed to whatever you configure when you bring it up. We're working on code to do all this at boot, but it's not ready yet.
Bill
On Wed, Mar 15, 2006 at 02:08:43PM -0500, Bill Nottingham wrote:
Julian C. Dunn (julian.dunn@devlin.ca) said:
You have a wired ethernet that is configured to be eth0, and your wireless module was loaded first, so it was moved out of the way.
Okay - is that expected behaviour? (I'm currently running without wireless configured, but I'm not suppressing the load of the kernel module or anything like that)
Yes. If you configre the wireless, it will be changed to whatever you configure when you bring it up. We're working on code to do all this at boot, but it's not ready yet.
At least in theory you should be able to give some "non-weird" and a fixed name to your wireless interface in /etc/mactab. Check 'man nameif' for a description. I did not test that but various scripts in /etc/sysconfig/network-scripts make an impression that they try to take /etc/mactab into consideration.
Surely you can get that if you will force a run of 'nameif' early enough in a boot sequence but this means some extra startup scripts and likely is not needed at all.
Michal
Michal Jaegermann (michal@harddata.com) said:
At least in theory you should be able to give some "non-weird" and a fixed name to your wireless interface in /etc/mactab. Check 'man nameif' for a description. I did not test that but various scripts in /etc/sysconfig/network-scripts make an impression that they try to take /etc/mactab into consideration.
Surely you can get that if you will force a run of 'nameif' early enough in a boot sequence but this means some extra startup scripts and likely is not needed at all.
nameif:
- doesn't handle collisions among names (just dies with -EEXIST) - fails really badly on devices that create both eth0 and wifi0 (can change the wrong device)
Bill
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
This just happened to me in a recent (last day or too) rawhide update.
It also appears that yum removed the kernel I was running, instead of the more recent kernel that wasn't currently running (and hadn't been used as I don't always get around to rebooting now that suspend/resume works so nicely.)
I'll give kernel versions as soon as I get rawhide running again.
R.
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
I too had this happen on my laptop. It only happened on one of my three rawhide systems so it's hard to figure out exactly why.
tjb -- ======================================================================= | Thomas Baker email: tjb@unh.edu | | Systems Programmer | | Research Computing Center voice: (603) 862-4490 | | University of New Hampshire fax: (603) 862-1761 | | 332 Morse Hall | | Durham, NH 03824 USA http://wintermute.sr.unh.edu/~tjb | =======================================================================
Thomas J. Baker wrote:
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
Yes, I did as well - I conjecture that the postinstall script of initscripts barfed, and thus the old inittab had been moved, but the new one not put into place. At least it isn't hard to fix *if* you know how to boot into single user mode from grub, remount root as read/write, and rename the file - in other words, a major disaster for a newbie, a minor hassle for an old pro.
David D. Hagood wrote:
Thomas J. Baker wrote:
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that
inittab
was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
Yes, I did as well - I conjecture that the postinstall script of initscripts barfed, and thus the old inittab had been moved, but the new one not put into place. At least it isn't hard to fix *if* you know how to boot into single user mode from grub, remount root as read/write, and rename the file - in other words, a major disaster for a newbie, a minor hassle for an old pro.
I had a report on this issue from one of the testers.
Rahul
On Sat, Oct 14, 2006 at 12:25:50PM -0400, Thomas J. Baker wrote:
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
I too had this happen on my laptop. It only happened on one of my three rawhide systems so it's hard to figure out exactly why.
Things like that happen when instead of updating a package yum decides to remove an old package first and later to install a new version. Look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=196590 for similar reports. Does it ring a bell? If you have seen something of that sort recently you likely should add some comments in bugzilla.
When I tried to repeat something like that I could not.
Michal
Michal Jaegermann wrote:
On Sat, Oct 14, 2006 at 12:25:50PM -0400, Thomas J. Baker wrote:
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel). When I rebooted, init complained that inittab was missing, so I rebooted into rescue mode and found that there was an /etc/inittab.rpmsave but no /etc/inittab. Has anyone else had this problem?
I too had this happen on my laptop. It only happened on one of my three rawhide systems so it's hard to figure out exactly why.
Things like that happen when instead of updating a package yum decides to remove an old package first and later to install a new version. Look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=196590 for similar reports. Does it ring a bell? If you have seen something of that sort recently you likely should add some comments in bugzilla.
When I tried to repeat something like that I could not.
Michal
What is happening here? Is the new package upgraded, then the older package removed, taking out files from the newly installed package? Or is the old package removed, then the new package performs an upgrade routine, which does not install the files in upgrade mode vs. install mode? Jim
On Sat, Oct 14, 2006 at 09:37:59PM -0400, Jim Cornette wrote:
Michal Jaegermann wrote:
On Sat, Oct 14, 2006 at 12:25:50PM -0400, Thomas J. Baker wrote:
On Wed, 2006-03-15 at 10:18 -0500, Julian C. Dunn wrote:
I just did a yum update, updating to initscripts 8.31.1-1 (among other things like a new kernel).
I too had this happen on my laptop. It only happened on one of my three rawhide systems so it's hard to figure out exactly why.
Things like that happen when instead of updating a package yum decides to remove an old package first and later to install a new version. Look at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=196590
What is happening here? Is the new package upgraded, then the older package removed, taking out files from the newly installed package?
No, not really. At least it does not seem to be that way. A bug report quoted above has examples where such problem was observed. Look at a sequence of events which follows "Running Transaction" to see "Removing/Updating" and effects.
If this would happen with packages which do not have config files then you would not get into any troubles . The issue appears to be that "remove" step saves "not-anymore-needed" configuration files saving them with .rpmsave suffix, but "update" knows that it should not clobber config files which originally were there so it refrains from unpacking defaults ones. That is actually what it _should_ do. Net effects may turn out to be nasty.
I assume that what Thomas and Julian are talking about is another manifestation of that bug. If that is really something else that would interesting too.
Michal