Beyond The Data
We hit Browser compatible issues on a lot of old EL5 Linux Environment.
An alternative is using Firefox ESR version.
Just download RPM. and issue #yum localinstall firefox*rpm
Ref:
http://rpmfind.net//linux/RPM/centos/updates/5.11/x86_64/RPMS/firefox-31.2.0-3.el5.centos.x86_64.html
Env:
Oracle VM Server 3.3.1
Host: hostbase0
VM: hostvm0
Problem:
User cannot access from outside cluster via ssh.
ssh will hang here:
|
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 |
[suse@suse ~]$ssh hostvm0 -l root -vvv ... ... debug2: kex_parse_kexinit: reserved 0 debug2: mac_setup: found hmac-md5 debug1: kex: server->client aes128-ctr hmac-md5 none debug2: mac_setup: found hmac-md5 debug1: kex: client->server aes128-ctr hmac-md5 none debug1: SSH2_MSG_KEX_DH_GEX_REQUEST(1024<1024<8192) sent debug1: expecting SSH2_MSG_KEX_DH_GEX_GROUP debug2: dh_gen_key: priv key bits set: 138/256 debug2: bits set: 482/1024 debug1: SSH2_MSG_KEX_DH_GEX_INIT sent debug1: expecting SSH2_MSG_KEX_DH_GEX_REPLY <-- stops here ^C |
|
1 2 3 4 5 6 7 8 9 10 11 12 13 |
#ssh host06 -vvv ... ... debug2: kex_parse_kexinit: first_kex_follows 0 debug2: kex_parse_kexinit: reserved 0 debug2: mac_setup: found hmac-md5 debug1: kex: server->client aes128-ctr hmac-md5 none debug2: mac_setup: found hmac-md5 debug1: kex: client->server aes128-ctr hmac-md5 none debug1: SSH2_MSG_KEX_DH_GEX_REQUEST(1024<1024<8192) sent debug1: expecting SSH2_MSG_KEX_DH_GEX_GROUP <--stops here. debug3: Wrote 24 bytes for a total of 1005 |
from ifconfig output, I can see a lot of err/invalid packages:
|
1 |
RX packets:1321 errors:21 dropped:1073 overruns:0 frame:0 |
from sys log: can see a lot of err mesg on xen_netfront.
|
1 2 3 4 5 |
Nov 5 21:32:18 hostvm0 kernel: xen_netfront: GSO size must not be zero Nov 5 21:32:20 hostvm0 kernel: xen_netfront: GSO size must not be zero Nov 5 21:32:51 hostvm0 kernel: xen_netfront: GSO size must not be zero Nov 5 21:32:57 hostvm0 kernel: xen_netfront: GSO size must not be zero Nov 5 21:32:57 hostvm0 kernel: xen_netfront: GSO size must not be zero |
Fix:
It’s an oracle bug which marked in Metalink ID:1929676.1
Upgrade Linux kernel on Host to latest:
current kernel: kernel-uek-3.8.13-26.4.2.el6uek.x86_64
|
1 2 3 4 5 |
# yum update -y kernel-uek ... ... Installing : kernel-uek-firmware-3.8.13-44.1.4.el6uek.noarch 1/2 Installing : kernel-uek-3.8.13-44.1.4.el6uek.x86_64 2/2 |
Env:
Oracle ODA Appliance, which based on ORACLE VM Server set.
Dom0/Host: odahost1
Dom1/VM: odahostdb1
Problem:
use can not ssh onto Dom1/VM from outside ODA. ping is ok.
SSH will hang here:
|
1 2 3 4 5 |
#ssh root@odahostdb1 ... ... debug1: expecting SSH2_MSG_KEX_DH_GEX_GROUP Connection closed by 10.191.6.55 |
From Dom0 /var/log/message:
|
1 2 3 4 5 |
Nov 21 21:40:49 odahost1 last message repeated 2 times Nov 21 21:41:50 odahost1 last message repeated 7 times Nov 21 21:44:40 odahost1 kernel: bond0: received packets cannot be forwarded while LRO is enabled Nov 21 21:46:48 odahost1 last message repeated 16 times Nov 21 21:48:48 odahost1 last message repeated 8 times |
This problem is due to LRO of eth0 not disabled.
Although bond0 LRO is disabled by default, eth0 which bond0 is based on is not.
|
1 2 3 4 |
[root@odahost1 ~]# ethtool -k bond0 | grep large-receive-offload large-receive-offload: off [root@odahost1 ~]# ethtool -k eth0 | grep large-receive-offload large-receive-offload: on |
Fix:
just disable LRO on eth0
|
1 |
[root@odahost1 network-scripts]# ethtool -K eth0 lro off |
restart network service.
To make the changes permenant:
If Dom0/Host is EL5:
add following line to /etc/rc.local
|
1 |
/sbin/ethtool -K eth0 lro off |
If Dom0/Host is EL6+:
add following line to ifcfg-eth0:
|
1 |
ETHTOOL_OPTS="-K ${DEVICE} lro off" |
the problem on EL5/6 different fix is on initscripts package version.
EL6 with initscripts 9.03+ accepts this ETHTOOL_OPTS, but EL5 with initscripts 8.45 does not.
Refer:
Poor TCP Performance on VMs; dmesg:”received packets cannot be forwarded while LRO is enabled”