Plesk 8.0.1 appears to trash boot config

[kurtiswalsh]

Installed clean version of Fedora Core 4 on Compaq DL360 server with hardware RAID 1. Fedora install went without issue and would boot fine. I downloaded auto installer for Fedora Core 4 and installed Plesk 8.0.1. Installed fine and was operational. When I rebooted the machine it gave me an error message about etc/fstab and the machine halted and would not continue to boot. It also complained about files being read-only. I'm assuming this is because the fstab didn't complete.

Anyone else seen this issue? Any suggestions?

Update:
After posting about the DL360 failing, I decided to try installing this on an ML370 server with a RAID 1 Mirror as root and boot and a RAID 5 partition as /var. Again Fedora core 4 installed with no issues and would boot up and was fully operational. Once I installed Plesk 8.0.1 on the machine and rebooted the machine comes up compaining about 'Line 8 in etc/fstab is bad'. It also is going through lists of files it can no longer find. The machine will no longer complete the booting process.

I'm new to Plesk and would like to get this installed on my systems, but with two machines down I'm starting to question if this product is right for managing servers. Perhaps its made to be installed on desktop PC's? Don't know.

Any suggestions?

Kurt

 

[tnats@]

I'm having simliar issues with RedHat Enterprise.

It says /var is gone in the bootup screen and it won't boot anymore.

 

[kurtiswalsh]

Originally posted by tnats
I'm having simliar issues with RedHat Enterprise.

It says /var is gone in the bootup screen and it won't boot anymore.

Have you had any previous versions of Plesk running? If so, what OS and what version of Plesk?

 

[tnats@]

Mine is a fresh install of RedHat Enterprise. I made / partition 6gig and /var 65gig.

I then installed Plesk. It said it installed correctly. I then rebooted and this is where the bootup console is saying /var doesn't exist anymore. (example: no such file /var/log/dmesg)

 

[kurtiswalsh]

Yup, we are having the same problem. Mine first complains about the fstab having an invalid line. On the DL360 it showed all the /var as read-only and could not change them to rw. On the ML370 it reports "file or directory not found" for a bunch of files. I think they are all on /var.

 

[tnats@]

I have submitted a ticket. We have email support. I'll keep this thread updated on my resolution. In the meantime, I'm reinstalling RedHat because there doesn't seem to be any way to recover it.

 

[tnats@]

*update*

I wiped out my install and just put everything under the / partition and reinstalled.

It looked good, both plesk and sitebuilder installed and I worked with them for a bit. I then rebooted and the box came back this time. Plesk seems to work fine but Sitebuilder gives this error: Fatal error: extension 'gd' was not found.

Also, as I'm sitting here, I'm getting these messages on my ssh console:
Message from syslogd@plesk1 at Thu Jun 22 22:55:29 2006 ...
plesk1 kernel: journal commit I/O error


That doesn't look good.....

Ah, the whole box has crashed. The drives are hosed and it's in read-only mode. Nice.

 

[tnats@]

Ok, I confirmed that it has trashed /var again. Same problem.

It says /var doesn't exist again.

 

[kurtiswalsh]

Good times! Any news from SWSoft? Are you installing on hardware RAID partitions?

I installed FreeBSD 6.1 on my machine last night and tried the auto-installer for 6.0, but no luck. The versions didn't match. BSD website does not show 6.0 being available. I'll have to do some more digging to find that and try it. Man, BSD installer is not that friendly.

 

[tnats@]

Trying to get them to help me.

I just re-installed RedHat for the 3rd time but I'm not going to do anything else right now until I can get them to help me.





I know it's a RedHat issue because last week, I installed Plesk on Debian on a hardware RAID box and it worked fine.

 

[Connor]

I had this problem with CentOS 4.3 with the system not finding the /var partiton.. On further review.. It looks as if diskquata is mucking up the fstab and adding a extra /var into and remove the 2 from the end of the fstab line.. to fix.. I booted off the CentOS disk and went into rescue mode.. chrooted the filesystem and fixed the fstab file manually.. rebooted, and everything is good.. That step can be avoided if you fix the fstab file BEFORE rebooting after you Upgrade to 8.0.1 Please note, I didn't have this problem when I installed 8.0.0, only after upgrading to 8.0.1 or doing a clean install with 8.0.1

 

[tnats@]

Thanks for the reply. Just got off the phone with them. They were thinking the same thing but he had me go look at my current fstab which is still correct.

Thanks for the post. They are in my box now doing the install.

 

[DataHive]

I suspect my Plesk install is doing the same thing. I ordered a new box from ev1servers to replace my current Plesk 5 box, and after a reboot Plesk will not start. I just get errors such as:

failed to make symlink /etc/rc2.d/S56xinetd: Read-only file system
failed to make symlink /etc/rc3.d/S56xinetd: File exists
failed to make symlink /etc/rc4.d/S56xinetd: File exists
failed to make symlink /etc/rc5.d/S56xinetd: File exists
Starting xinetd: [ OK ]
touch: cannot touch `/var/lock/subsys/xinetd': Read-only file system
failed to make symlink /etc/rc2.d/S55named: File exists
failed to make symlink /etc/rc3.d/S55named: File exists
failed to make symlink /etc/rc4.d/S55named: File exists
failed to make symlink /etc/rc5.d/S55named: File exists
Starting named: cp: cannot remove `/var/named/run-root/etc/localtime': Read-only file system
touch: cannot touch `/var/lock/subsys/named': Read-only file system

[ errors continue for a few screens ]

I've logged a support request with SWsoft.

Sam

 

[tnats@]

Here is my fstab. They had to add the usrquota in there.

My /var is under / so that is where it is.

[root@plesk1 mail]# more /etc/fstab
# This file is edited by fstab-sync - see 'man fstab-sync' for details
/dev/VolGroup00/LogVol00 / / ext3,usrquota defaults 1
LABEL=/boot /boot ext3 defaults 1 2
none /dev/pts devpts gid=5,mode=620 0 0
none /dev/shm tmpfs defaults 0 0
none /proc proc defaults 0 0
none /sys sysfs defaults 0 0
/dev/VolGroup00/LogVol01 swap swap defaults 0 0
/dev/hda /media/cdrom auto pamconsole,fscontext=system_ubject_r:remov
able_t,exec,noauto,managed 0 0
/dev/fd0 /media/floppy auto pamconsole,fscontext=system_ubject_r:remov
able_t,exec,noauto,managed 0 0

 

[DataHive]

Doesn't mean much to me. I'm still waiting to hear from Plesk, so hopefully they'll have something soon.

 

[DataHive]

Oh well, SWsoft sent me back to ev1servers, unless I pony up a support fee.

Ev1 have recommended starting from scratch with a fresh server.

Thankfully I only put a days work into the server so far *groan*.

Given how little I did to the server to make it break, i'm kinda thinking it'll just happen again. I guess i'll just hammer it as hard as possible straight up, and see if I can break it again.

Sam

 

[DataHive]

Further investigation:

As mentioned by someone else earlier, it looks like the /etc/fstab has been munged in some way. The first line in my fstab file is:

/dev/hda5 / / ext3,usrquota defaults 1

mount doesn't like it:

[mntent]: line 2 in /etc/fstab is bad
mount: can't find /dev/hda5 in /etc/fstab or /etc/mtab

As a result, the majority of the filesystem has been mounted as read-only, and hence nothing works properly.

The question is, how the hell can I fix fstab when it currently resides on a read-only filesystem?

Sam

 

[DataHive]

And as far as I can gather, that first line in the fstab file should be something more like:

/dev/hda5 / ext3 usrquota,defaults 1 2

Sam

 

[tnats@]

I posted my fstab on the previous page. If it's munged, I can email it to you. Here is what I would do. Re-install the OS then install plesk, then before you reboot, fix the fstab.

I'm trying to get clarifcation on what swsoft did in order to get my install to work.

 

[DataHive]

Yeah, the re-install may be the only option, seeing as I don't have emergency access to the box to get that filesystem mounted, and the file fixed up. ev1 are currently working on either fixing that file, or re-imaging the server.