Xen kernel help

dnom

I have this xen vps that was recently migrated and didn't properly boot. Using the VNC console I was able to login to a read only filesystem. I saw that my storage drive /dev/xvda1 was missing but I was able to find a /dev/sda1 with the same size. I went and try to boot to recovery, did fsck and found no issues, mounted the drive (the recovery OS sees the storage drive as /dev/xvda1), and changed my /etc/fstab to mount /dev/sda1 instead and I was able to boot normally.

I still had an issue with ssh not working so I did some googling with the error and was able to fix it. Now I am having issue starting up apache with this error message:
[crit] (22)Invalid argument: alloc_listener: failed to get a socket for (null) I haven't checked what kernel I was using previously because everything just works but all the problems I encountered tells me that I am now using a different kernel than what I was using previously.

uname -r shows:

2.6.18-348.12.1.el5xen

Is there anyway I can fix this myself? I am on Xen PV so I couldn't just install a newer kernel (or maybe I actually can?)

dnom

Noone?

Bumping this just in case our Xen experts weren't around when I posted and it sunk down so fast.

To mods: If this is against the rules, pls delete.

racksx

Have you try upgrade your apache version? i had this issue before, and it was related to apache lib-apr

AnthonySmith

Your OS is centos 5 then I assume?

If you can boot in to recovery then you need to change the devicemap in /boot/grub to sda instead of xvda also look in your /boot/grub/grub.conf and make sure and references to xvda are changed to sda

Really this is not something you should fix yourself, it's the config file on the host specifying the wrong disk type.

Iirc centos 5 is the only OS left using sda instead of xvda or xvde thankfully it will be end of life in about 5 minutes nth so or so.

Edit: the above is only relevant if your host is running pygrub looking at that again it seems your host might not be, if your sharing the host kernel then you need to change hosts asap, the node is so out of date it's almost criminal.

Which host is this out of interest?

Edit 2: sorry just retread that missed the second paragraph completely!

You should be able to upgrade your kernel unless your not on pygrub and the hostnode is running a VERY old Xen version and kernel.

dnom

I'm on CentOS 6.8. I have the latest kernel installed my grub.conf looks like this:

default=0
timeout=5
title CentOS (2.6.32-642.4.2.el6.x86_64)
  root (hd0,0)
  kernel /boot/vmlinuz-2.6.32-642.4.2.el6.x86_64 console=hvc0 xencons=tty0 root=/dev/xvda1 ro crashkernel=auto LANG=en_US.UTF-8
title CentOS (2.6.32-573.8.1.el6.x86_64)
  root (hd0,0)
  kernel /boot/vmlinuz-2.6.32-573.8.1.el6.x86_64 console=hvc0 xencons=tty0 root=/dev/xvda1 ro crashkernel=auto LANG=en_US.UTF-8
  initrd /boot/initramfs-2.6.32-573.8.1.el6.x86_64.img
title CentOS (2.6.32-573.7.1.el6.x86_64)
  root (hd0,0)
  kernel /boot/vmlinuz-2.6.32-573.7.1.el6.x86_64 console=hvc0 xencons=tty0 root=/dev/xvda1 ro crashkernel=auto LANG=en_US.UTF-8
  initrd /boot/initramfs-2.6.32-573.7.1.el6.x86_64.img
title vmlinuz-2.6.32-431.el6.x86_64
  root (hd0,0)
  kernel /boot/vmlinuz-2.6.32-431.el6.x86_64 console=hvc0 xencons=tty0 root=/dev/xvda1 ro
  initrd /boot/initramfs-2.6.32-431.el6.x86_64.img

but uname -r shows the old el5xen kernel

dnom

This is on a drserver xen vps that had a lot trouble recently. I don't really have any important files here(I didn't even notice it wasn't up until recently) but I just want to try fix things myself if I can whenever anything breaks on my servers and learn a little bit more on troubleshooting.

AnthonySmith

right, so 2.6.18-348.12.1.el5xen was just from the rescue environment, dair enough.

I thought Dr.Server only uses Xen HVM? either way this has nothing to do with the host node at all.

Try a complete reinstall of apache for a start.

dnom

No 2.6.18-348.12.1.el5xen is on the main system. The recovery environment has 3.16.0-3-amd64

$ virt-what
xen
xen-dom

still having the same problem after apache reinstall and nginx is also having problem. I don't think I should be using 2.6.18-348.12.1.el5xen on centos 6.8 and it's not listed on my grub.conf so I'm not really sure what's going on.

AnthonySmith

Well the only reason would be that they have centos 5 on the host node and your are on xen pv and they don't have pygrub enabled for your server, you cant fix that as it is pre boot from your perspective, it is a 2 second fix, you need to pop a ticket in.

dnom

AnthonySmith said: Well the only reason would be that they have centos 5 on the host node and your are on xen pv and they don't have pygrub enabled for your server, you cant fix that as it is pre boot from your perspective, it is a 2 second fix, you need to pop a ticket in.

Case closed then.

Thanks!!

Ikoula

It is probably a litlle late and maybe i am wrong but your issue makes me think about a pv driver issue (also known as xentools).

edit : maybe the pv driver of your vm is older than the one used on the new host.

Radi

@AnthonySmith He is on PV, and we have CentOS 6.6 on that node. Enabling pygrub as we speak.

dnom

I'm now able to boot with the installed kernel, @Radi enabled pygrub but it still didn't boot at first. I was able to get into the vnc console for a few seconds after powering it off before the vnc console closes down. I manage to get a screenshot and found out that my volume device name changed again, this time to "xvde1" so I manually changed my grub.conf and /etc/fstab file.

Apache is now starting up without issue

Thanks!