There are some bugs in systemd, plymouth and maybe dracut that will make it very difficult to get your system rebooted if you have an encrypted home (or other non-root) file system. It's possible this will also be the case now even if you don't have encrypted partitions. Currently when the boot is failing, systemd appears to be confused about the state of the system and writes some files (/etc/profile and /etc/fstab) as though the root pivot hadn't happened yet. The git3 kernel was crashing and booting using an encypted root has worked for initramsfs images built within the last week. So there isn't a lot of incentive to reboot to test new kernels right now anyway.
No freakin' kidding :(
On Sun, Jul 8, 2012 at 11:29 AM, Bruno Wolff III bruno@wolff.to wrote:
There are some bugs in systemd, plymouth and maybe dracut that will make it very difficult to get your system rebooted if you have an encrypted home (or other non-root) file system. It's possible this will also be the case now even if you don't have encrypted partitions. Currently when the boot is failing, systemd appears to be confused about the state of the system and writes some files (/etc/profile and /etc/fstab) as though the root pivot hadn't happened yet. The git3 kernel was crashing and booting using an encypted root has worked for initramsfs images built within the last week. So there isn't a lot of incentive to reboot to test new kernels right now anyway. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.**org/mailman/listinfo/testhttps://admin.fedoraproject.org/mailman/listinfo/test
Is there a suggested recovery strategy for backing out to a bootable state?
Thanks :)
On Sun, Jul 8, 2012 at 2:39 PM, Roger Barraud roger.barraud@gmail.comwrote:
No freakin' kidding :(
On Sun, Jul 8, 2012 at 11:29 AM, Bruno Wolff III bruno@wolff.to wrote:
There are some bugs in systemd, plymouth and maybe dracut that will make it very difficult to get your system rebooted if you have an encrypted home (or other non-root) file system. It's possible this will also be the case now even if you don't have encrypted partitions. Currently when the boot is failing, systemd appears to be confused about the state of the system and writes some files (/etc/profile and /etc/fstab) as though the root pivot hadn't happened yet. The git3 kernel was crashing and booting using an encypted root has worked for initramsfs images built within the last week. So there isn't a lot of incentive to reboot to test new kernels right now anyway. -- test mailing list test@lists.fedoraproject.org To unsubscribe: https://admin.fedoraproject.**org/mailman/listinfo/testhttps://admin.fedoraproject.org/mailman/listinfo/test
-- Regards Roger Barraud Auckland New Zealand
On Sun, Jul 08, 2012 at 15:03:35 +1200, Roger Barraud roger.barraud@gmail.com wrote:
Is there a suggested recovery strategy for backing out to a bootable state?
I tried doing it on one machine but didn't have much luck because the minium requirements for several packages all changed at once and the old udev stuff was obsoleted. I managed to toast the root file system on one machine, but I don't think that was directly caused by the bugs.
If you roll back everything to when the rc5 git0 kernel was built, things should work except for encrypted home and you can still mount home (and encrypted swap) if need be and then start graphical.target.
If you have an older initramfs image saved you probably don't need to go back quite that far, as I don't think things got really broken until until Tuesday. With systemd-185-7.gite7aee75.fc18 you can probably boot again. However I think the update following that (systemd-186-1.fc18) was the one that obsoleted a udev package. I remember trying to go back and having troubles getting everything I needed. But I had already toasted a chunk of the root file system by then and am not sure which issue was blocking things.
I have been thinking about how I might try to handle recovering my other rawhide system, but was really hoping there'd be a fixed systemd I could update to, rather than try a complicated downgrade.
I am going to use an f15 live image to mount my file systems and chroot so that I can use yum to do upgrades or downgrades as needed.
On Sat, Jul 07, 2012 at 23:44:27 -0500, Bruno Wolff III bruno@wolff.to wrote:
I am going to use an f15 live image to mount my file systems and chroot so that I can use yum to do upgrades or downgrades as needed.
I forgot to mention you are probably going to need to remove the sysroot lines that get added to /etc/fstab before manually starting the graphical.target. Once you are back up you'll want to get a fresh copy of /etc/profile. You can remove /etc/profile and reinstall setup to get it back to the default.
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
1. Reinstalled F17 LXDE spin 2. Update everything except kernel and polkit to Rawhide
Crude but effective -- Steven I Usdansky, PhD Traveling Geologist
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
Il giorno dom, 08/07/2012 alle 19.56 -0400, David ha scritto:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
Yes, and I can assure you (after I spent ~6 hours fixing the mess of the last upgrade in a sane way, downgrading packages by hand), that Fedora Rawhide with encrypted partitions will not be able to mount the root partition.
Cheers, Matteo
On 07/08/2012 04:56 PM, David wrote:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
I did not ask for any encrypted file system. It just didn't work.
On 7/8/2012 9:19 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 04:56 PM, David wrote:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
I did not ask for any encrypted file system. It just didn't work.
The last time I had anything to do with "any encrypted file system" and Fedora you had to actually ask for it to be encrypted with the install. It was not the default. So? My thoughts are that you asked for and had one before now.
Your turn.
On 07/08/2012 06:54 PM, David wrote:
On 7/8/2012 9:19 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 04:56 PM, David wrote:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote:
After upgrading my Rawhide installation yesterday, I couldn't even
reboot using an older kernel, so I got out the proverbial hammer and:
- Reinstalled F17 LXDE spin
- Update everything except kernel and polkit to Rawhide
Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
I did not ask for any encrypted file system. It just didn't work.
The last time I had anything to do with "any encrypted file system" and Fedora you had to actually ask for it to be encrypted with the install. It was not the default. So? My thoughts are that you asked for and had one before now.
Your turn.
I did not ask for an encrypted file system. I have never used an encrypted file system with Linux.
On 7/8/2012 10:01 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 06:54 PM, David wrote:
On 7/8/2012 9:19 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 04:56 PM, David wrote:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote:
On Sun, Jul 08, 2012 at 03:54:50 -0700, Steven I Usdansky usdanskys@rocketmail.com wrote: > After upgrading my Rawhide installation yesterday, I couldn't even > > reboot using an older kernel, so I got out the proverbial hammer > and: > > 1. Reinstalled F17 LXDE spin > 2. Update everything except kernel and polkit to Rawhide > > Crude but effective
I'm going to try revert dracut, plymouth, udev and systemd to the pacakges from around May 24th. I won't get a chance to try this until late tonight. I'll report back after giving this a try.
I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
I did not ask for any encrypted file system. It just didn't work.
The last time I had anything to do with "any encrypted file system" and Fedora you had to actually ask for it to be encrypted with the install. It was not the default. So? My thoughts are that you asked for and had one before now.
Your turn.
I did not ask for an encrypted file system. I have never used an encrypted file system with Linux.
Okay. Since I have never seen anyone say that they encrypted their file system using Fedora *after* the install an you now say that you file system is encrypted?
I can offer no help or suggestions.
This looks like another one of your many self-inflected distastes reported over many lists, over many years.
Good luck with this. I'm gone.
On 07/08/2012 07:18 PM, David wrote:
On 7/8/2012 10:01 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 06:54 PM, David wrote:
On 7/8/2012 9:19 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 04:56 PM, David wrote:
On 7/8/2012 4:37 PM, Chuck Forsberg WA7KGX N2469R wrote:
On 07/08/2012 10:12 AM, Bruno Wolff III wrote: > On Sun, Jul 08, 2012 at 03:54:50 -0700, > Steven I Usdansky usdanskys@rocketmail.com wrote: >> After upgrading my Rawhide installation yesterday, I couldn't even >> >> reboot using an older kernel, so I got out the proverbial hammer >> and: >> >> 1. Reinstalled F17 LXDE spin >> 2. Update everything except kernel and polkit to Rawhide >> >> Crude but effective > > I'm going to try revert dracut, plymouth, udev and systemd to the > pacakges from around May 24th. I won't get a chance to try this > until > late tonight. I'll report back after giving this a try. I made another attempt at installing with last week's netinst. This time I used defaults wherever possible. The result was the same. An installation dialog resembling Suse that installs a Grub that complains about a missing splash file and can't find much of anything.
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
You guys have something really wrong. I do not have these problems.
Did I understand correctly that you have encrypted / directories. That would be my first 'I don't have that' comment.
I did not ask for any encrypted file system. It just didn't work.
The last time I had anything to do with "any encrypted file system" and Fedora you had to actually ask for it to be encrypted with the install. It was not the default. So? My thoughts are that you asked for and had one before now.
Your turn.
I did not ask for an encrypted file system. I have never used an encrypted file system with Linux.
Okay. Since I have never seen anyone say that they encrypted their file system using Fedora *after* the install an you now say that you file system is encrypted?
I can offer no help or suggestions.
This looks like another one of your many self-inflected distastes reported over many lists, over many years.
Good luck with this. I'm gone.
I just checked the root file system installed by the July 3 netinstall using a rescue CD. The file system was not encrypted. That is reasonable as I did not ask for an encrypted file system. I have never knowingly installed an encrypted file system. They're flakey enough without encryption. Neither have I ever converted a file system to encrypted. Whoever thought I said I used an encrypted file system needs some remedial reading comprehension lessons.
On Sun, Jul 08, 2012 at 18:19:53 -0700, Chuck Forsberg WA7KGX N2469R caf@omen.com wrote:
I did not ask for any encrypted file system. It just didn't work.
It looks like the current issue is likely fallout from installing dractut-020-51 as noted in one of the other replies. This probably breaks things whether or not you are using encrypted file systems. There appears to be two other bugs related to encrtped file systems. One is that plymouth didn't get the systemd units to ask for a password for file systems mounted after the root pivot (e.g. swap and home). The other seems to affected systems with encrypted root.
On Mon, Jul 09, 2012 at 07:08:06 -0500, Bruno Wolff III bruno@wolff.to wrote:
On Sun, Jul 08, 2012 at 18:19:53 -0700, Chuck Forsberg WA7KGX N2469R caf@omen.com wrote:
I did not ask for any encrypted file system. It just didn't work.
It looks like the current issue is likely fallout from installing dractut-020-51 as noted in one of the other replies. This probably breaks things whether or not you are using encrypted file systems. There appears to be two other bugs related to encrtped file systems. One is that plymouth didn't get the systemd units to ask for a password for file systems mounted after the root pivot (e.g. swap and home). The other seems to affected systems with encrypted root.
After cleaning up from dracut-020-51, installing dracut-020-57.git20120709.fc18 and running the kernel scripts (to run dracut) I can now boot. Don't the plymouth bug is still there, so I have to manually deal with home, swap and then starting the graphical.target.
I also am running a slightly older kernel. kernel-3.5.0-0.rc5.git3.1.fc18 crashes very early in the boot process.
On Mon, Jul 09, 2012 at 08:09:08 -0500, Bruno Wolff III bruno@wolff.to wrote:
After cleaning up from dracut-020-51, installing dracut-020-57.git20120709.fc18 and running the kernel scripts (to run dracut) I can now boot. Don't the plymouth bug is still there, so I have to manually deal with home, swap and then starting the graphical.target.
To actually get a graphical login I had to set enforcing mode to permissive. I am not sure if this is a bug or a labelling issue. (I'm relabelling now.) sedispath was getting blocked was tying up my disk IO bandwidth updating the audit file.
On Mon, Jul 9, 2012 at 3:35 PM, Bruno Wolff III bruno@wolff.to wrote:
On Mon, Jul 09, 2012 at 08:09:08 -0500,
Bruno Wolff III bruno@wolff.to wrote:
After cleaning up from dracut-020-51, installing dracut-020-57.git20120709.fc18 and running the kernel scripts (to run dracut) I can now boot. Don't the plymouth bug is still there, so I have to manually deal with home, swap and then starting the graphical.target.
To actually get a graphical login I had to set enforcing mode to permissive. I am not sure if this is a bug or a labelling issue. (I'm relabelling now.) sedispath was getting blocked was tying up my disk IO bandwidth updating the audit file.
I also have to use permissive mode, relabelling did not help.
On Mon, Jul 09, 2012 at 15:45:26 +0200, Sandro Mani manisandro@gmail.com wrote:
On Mon, Jul 9, 2012 at 3:35 PM, Bruno Wolff III bruno@wolff.to wrote:
To actually get a graphical login I had to set enforcing mode to permissive. I am not sure if this is a bug or a labelling issue. (I'm relabelling now.) sedispath was getting blocked was tying up my disk IO bandwidth updating the audit file.
I also have to use permissive mode, relabelling did not help.
I opened the following bug: https://bugzilla.redhat.com/show_bug.cgi?id=838574
If it seems to fit what you are seeing, you might want to add information to it.
On Sun, 2012-07-08 at 13:37 -0700, Chuck Forsberg WA7KGX N2469R wrote:
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
That's not how Rawhide works. The images in the Rawhide tree are automatically generated. There's no testing or release process. They just get built periodically. If they work, great. If they don't, no-one guaranteed that they would.
Dne 11.7.2012 22:18, Adam Williamson napsal(a):
On Sun, 2012-07-08 at 13:37 -0700, Chuck Forsberg WA7KGX N2469R wrote:
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
That's not how Rawhide works. The images in the Rawhide tree are automatically generated. There's no testing or release process. They just get built periodically. If they work, great. If they don't, no-one guaranteed that they would.
Which is pretty bad plan...
You want to have Rawhide being used by skilled people - to catch bugs early, not just a week before new Fedora release.
But if the quality of Rawhide will go to the road of trashing people's hard drives - skilled developers will leave Rawhide and will go for another distro.
I just hope there is minority of Fedora people who believe this is the right thing to do....
IMHO every package maintainer should seriously care about its package and always test it himself FIRST and avoid releasing packages with obvious killer bugs.
Zdenek
On 07/12/2012 02:53 AM, Zdenek Kabelac wrote:
Dne 11.7.2012 22:18, Adam Williamson napsal(a):
On Sun, 2012-07-08 at 13:37 -0700, Chuck Forsberg WA7KGX N2469R wrote:
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
That's not how Rawhide works. The images in the Rawhide tree are automatically generated. There's no testing or release process. They just get built periodically. If they work, great. If they don't, no-one guaranteed that they would.
Which is pretty bad plan...
You want to have Rawhide being used by skilled people - to catch bugs early, not just a week before new Fedora release.
But if the quality of Rawhide will go to the road of trashing people's hard drives - skilled developers will leave Rawhide and will go for another distro.
I just hope there is minority of Fedora people who believe this is the right thing to do....
IMHO every package maintainer should seriously care about its package and always test it himself FIRST and avoid releasing packages with obvious killer bugs.
Zdenek
I tend to agree with Kabelac. There has to be *some* testing that the developer is doing. He/she can't just throw in a bunch of new code and then let 'er rip, right?
Kevin
On Thu, 2012-07-12 at 09:53 +0200, Zdenek Kabelac wrote:
Dne 11.7.2012 22:18, Adam Williamson napsal(a):
On Sun, 2012-07-08 at 13:37 -0700, Chuck Forsberg WA7KGX N2469R wrote:
Before the next netinst is released perhaps the developers could be bothered to make sure it actually works.
That's not how Rawhide works. The images in the Rawhide tree are automatically generated. There's no testing or release process. They just get built periodically. If they work, great. If they don't, no-one guaranteed that they would.
Which is pretty bad plan...
You want to have Rawhide being used by skilled people - to catch bugs early, not just a week before new Fedora release.
But if the quality of Rawhide will go to the road of trashing people's hard drives - skilled developers will leave Rawhide and will go for another distro.
I just hope there is minority of Fedora people who believe this is the right thing to do....
IMHO every package maintainer should seriously care about its package and always test it himself FIRST and avoid releasing packages with obvious killer bugs.
You don't really need to be releasing packages with 'obvious killer bugs' to have issues generating network install images that work. The whole stack - _any_ whole stack of an OS all the way up to an interactive installer - is monstrously complicated and the only way to be really sure that complete generated installer images are viable and working is, well, to generate them and test each one, automatically or manually. It's really not plausible to expect every packager to test every possible use case of their package before updating it in Rawhide. Rawhide is a dev branch. There's a limit to how many testing requirements you can impose on a dev branch before it stops serving its purpose.
Note that nothing discussed in this thread has anything to do with 'trashing people's hard drives'. One bug has been discussed which breaks boot with encrypted partitions; then Chuck asserted that the auto-generated Rawhide netinst image currently isn't working. There was no report of any kind of data loss. Data loss bugs usually are treated with high urgency even in Rawhide.
I don't think it's realistically possible to expect the dev branch of an entire Linux distribution which is fast-moving, maintained by a fairly small group of developers, and has fairly limited QA resources to be reliably functional all the time. We can aim to _improve_ things, of course, we always can, but we have to set realistic goals...
When can we expect a netinst that actually installs a Linux that boots into something that looks like a working Linux installation?
I would prefer to be able to install F18 from a local tree via PXEBOOT but a quadratic internet netinst that works would be better than nothing.
On Thu, Jul 12, 2012 at 12:54:58 -0700, Chuck Forsberg WA7KGX N2469R caf@omen.com wrote:
When can we expect a netinst that actually installs a Linux that boots into something that looks like a working Linux installation?
I would prefer to be able to install F18 from a local tree via PXEBOOT but a quadratic internet netinst that works would be better than nothing.
I used the F17 installer to install F18 after trashing my rootfs last week. You can drop the updates repo and change the Fedora F17 repo to point to rawhide and do the install.
On 08/07/12 00:29, Bruno Wolff III wrote:
https://lists.fedoraproject.org/pipermail/devel/2012-July/169602.html
On Mon, Jul 09, 2012 at 09:30:08 +0100, Frank Murphy frankly3d@gmail.com wrote:
On 08/07/12 00:29, Bruno Wolff III wrote:
https://lists.fedoraproject.org/pipermail/devel/2012-July/169602.html
That appears to be related to the latest problem I was seeing. I'll try fixing that and see if I can get back to the previous errors.