Yum Ugrade 5.5 Centos - breaks networking?

J

jdexter

New Member
Nov 16, 2010
8
0
1
Halifax Nova Scotia
pve details:

prox1:~# pveversion -v
pve-manager: 1.7-10 (pve-manager/1.7/5323)
running kernel: 2.6.32-3-pve
proxmox-ve-2.6.32: 1.7-30
pve-kernel-2.6.32-3-pve: 2.6.32-14
pve-kernel-2.6.32-4-pve: 2.6.32-30
qemu-server: 1.1-28
pve-firmware: 1.0-10
libpve-storage-perl: 1.0-16
vncterm: 0.9-2
vzctl: 3.0.24-1pve4
vzdump: 1.2-10
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.13.0-3
ksm-control-daemon: 1.0-4


Set up default container (5.2) and then hit yum upgrade ... Now can't seem to get anywhere outside basic rack.. No DNS.. Seeming able to tracert to anything sharing IP block (10.10.130.*) .. but nothing else (name server 10.10.10.10 un-reachable) ..

Have I buggered the process by upgrade?

Jon
 
just search the forum for a solution, as far as I remember this is a known issue with centos.
 
Thx for getting back to me so quickly Tom .. Any hints? I see some stuff regarding 1.6 and regression - but - nothing on 1.7 and/or details on the fix..

Would you advise regressing the Proxmox install or is this related to centos across platforms? I tried to pull down one of the advanced centos containers from openVZ and it wouldn't let me use it. Listed as '-unknown-' in prox web interface.
 
Thx Tom - agreed the search is my friend .. :) ..

The problem is understanding which advise is correct on the problem (with thx for your links).

We are hoping to put ProxMox through some real paces here on a daily newspaper site with a stack of weird legacy configs PECL etc, Pressflow (drupal) and varnish/squid caching. Being stuck on genesis container is driving me nutz!

I'm pulling down:

http://download.openvz.org/template/precreated/contrib/centos-5-i386-default-5.5-20101104.tar.gz

.. and will try renaming.. Not completely understanding convention details, but will experiment using this file ..

Thx again for your quick assistance - will gladly write up case details as we get further along and throw some real load at it (about .5 million page transactions a day, with a peak around 170,000 an hour at 9:am) ..

Cheers - Jon
 
I just downloaded the template from the provided link, renamed it to centos-5-standard_5.5-2_i386.tar.gz and fired it up using venet. seems to work here. yum upgrade just downloaded 12 packages, poweroff and started again, still working.

another questions: what do you get on the host if you run ''pveperf'?
(a very simple benchmark tool).
 
Tried to wget directly into /var/lib/vz/template/cache - renamed as 'centos-5-standard_5.5-1_i386.tar.gz' .. still '-unkown-' in list.

So - renamed as above locally and am trying to upload via prox web interface now .. Bombed on first try - will try again..

I'm trying to upload as appliance template - should I upload as ISO?

Here is the output requested ..

output of pveperf (while uploading renamed container)

prox1:/var/lib/vz/template/cache# pveperf
CPU BOGOMIPS: 15960.22
REGEX/SECOND: 662946
HD SIZE: 19.69 GB (/dev/md1)
BUFFERED READS: 9.57 MB/sec
AVERAGE SEEK TIME: 4.37 ms
FSYNCS/SECOND: 35.92
DNS EXT: 62.61 ms
DNS INT: 9.27 ms (donair.org)
prox1:/var/lib/vz/template/cache# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/md1 20G 1.3G 18G 7% /
tmpfs 5.9G 0 5.9G 0% /lib/init/rw
udev 10M 584K 9.5M 6% /dev
tmpfs 5.9G 0 5.9G 0% /dev/shm
/dev/md4 114G 2.5G 111G 3% /var/lib/vz
 
Last edited:
jdexter said:
Tried to wget directly into /var/lib/vz/template/cache - renamed as 'centos-5-standard_5.5-1_i386.tar.gz' .. still '-unkown-' in list.
Click to expand...

'unkown' is ok. it should be usable via the gui but there is just no description about this template as this info is only available for our own templates (we maintain this file for our own templates)

jdexter said:
So - renamed as above locally and am trying to upload via prox web interface now .. Bombed on first try - will try again..

I'm trying to upload as appliance template - should I upload as ISO?

Here is the output requested ..

output of pveperf (while uploading renamed container)

prox1:/var/lib/vz/template/cache# pveperf
CPU BOGOMIPS: 15960.22
REGEX/SECOND: 662946
HD SIZE: 19.69 GB (/dev/md1)
BUFFERED READS: 9.57 MB/sec
AVERAGE SEEK TIME: 4.37 ms
FSYNCS/SECOND: 35.92
DNS EXT: 62.61 ms
DNS INT: 9.27 ms (donair.org)
Click to expand...

pls do not run pveperf when the server is under load. but you got extremly slow FSYNCS/SECOND, you should get have more than 1000. a reasonable server with a raid controller and BBU with write cache will give you much more.

BTW, And you are using a software raid config, which is not supported. But this should not be related to your template issues.
Do our templates work? e.g. our Debian 5?

jdexter said:
prox1:/var/lib/vz/template/cache# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/md1 20G 1.3G 18G 7% /
tmpfs 5.9G 0 5.9G 0% /lib/init/rw
udev 10M 584K 9.5M 6% /dev
tmpfs 5.9G 0 5.9G 0% /dev/shm
/dev/md4 114G 2.5G 111G 3% /var/lib/vz
Click to expand...
 
Hah !

Was just about to post back that while it is showing unknown - it is available in pull down under 'create' virtual machines ..

So - is there a running problem setting up as bridged ethernet rather than virtual network?

To be honest, not sure what the impact is - but set up bridged on the OEM container.

Should I not set this OpenVZ (Centos 5.5) container up bridged?

Cheers and again - thx!

Jon
 
See http://wiki.openvz.org/Differences_between_venet_and_veth

venet settings are done outside the container. this means you can set the IP via the proxmox ve gui. this is default and recommended.
if you go for veth, you will see a eth0 inside your container, which is NOT configured. you need to open a VNC console and create the needed IP configuration. On Debian you need to add it in /etc/network/interface, centos have different files for the network config.
 
yes, for centos, in /etc/sysconfig/network-scripts, add for example ifcfg-eth0, looking like:

DEVICE=eth0
BOOTPROTO=none
ONBOOT=yes
IPADDR=10.0.0.7
NETMASK=255.255.255.0
GATEWAY=10.0.0.254

(obviously change as you need it.)
 
Excellent - just learned several new things !!

Going to keep pounding at this and see what else I can bugger along the way .. I'm oddly, annoyingly, good at that ..

Promise I'll report in/ask more questions as we get deeper in.

We push our servers *very* hard and we are going to find out how ProxMox scales on multi-container/multi-node basis to address that demand.

Thanks again for the support/advise/patience - top notch and very helpful!

Cheers and best wishes.

Jon
 
/etc/resolv.conf, probably yeah. don't think so on the other one (don't remember having to fiddle with it...)
 
Hey !! that seems to have brought the beast alive!

This started with a yum upgrade - so upon checking - these files seemed to have already been set up correctly..

SO - now I'll try to restart, then shutdown and start the container and see if it comes back up..

Coolio !
 
restart inside the container will power off and it will take 5 minutes to start again, a cron job will start it (this is openvz specific).
 
You must log in or register to reply here.
Top Bottom