Service Guard and Apache

[MEHEE]

Hi,

I have some problem with the apache toolkit.

The error message :
###### Node "fep01.alpes.hp.com": Starting Package at Thu May 18 00:46:42 PDT 2006 ######
Incorrect metadata area header checksum
May 18 00:46:42 - Node "fep01.alpes.hp.com": Activating volume group vgpkgA .
May 18 00:46:42 - Node "fep01.alpes.hp.com": Checking filesystems:
/dev/vgpkgA/lvol0
e2fsck 1.35 (28-Feb-2004)
/dev/vgpkgA/lvol0: clean, 330/53772288 files, 1719636/107529216 blocks
May 18 00:46:42 - Node "fep01.alpes.hp.com": Mounting /dev/vgpkgA/lvol0 at /mnt/shared
May 18 00:46:42 - Node "fep01.alpes.hp.com": Adding IP address 192.168.0.97 to subnet 192.168.0.0
May 18 00:46:42 - Node "fep01.alpes.hp.com": Starting Apache daemons.
ERROR: Function start_server: Failed to start Apache daemons.
ERROR: May 18 00:46:42 - Failed to start HTTP.
ERROR: Function verify_ha_server; Failed to start HA servers
May 18 00:46:42 - Node "fep01.alpes.hp.com": Remove IP address 192.168.0.97 from subnet 192.168.0.0
May 18 00:46:42 - Node "fep01.alpes.hp.com": Unmounting filesystem on /mnt/shared
May 18 00:46:42 - Node "fep01.alpes.hp.com": Deactivating volume group vgpkgA
###### Node "fep01.alpes.hp.com": Package start FAILED at Thu May 18 00:46:42 PDT 2006 ######

My httpd.conf:
...
Listen 192.168.0.97
...

My control.sh:
...
IP[0]="192.168.0.97"
SUBNET[0]="192.168.0.0"
...
HA_APP_SERVER="post-IP"
...

I don't understand the problem with this configuration.
When i remove HA_APP_SERVER="post-IP" and start the "httpd -d $SERVER_ROOT -c "PidFile $PID_FILE" `get_moduleargs`" ( found in the hahttp.sh) in a shell, it's ok.

Somebody has an idea ?

Thanks,
MEHEE

[Steven E. Protter]

Shalom MEHEE,

Couple of possibilities.

You have a floating IP address configured with apache, perhaps there is a conflict with one of the other ip addresses ont he machine.

It would be helpful to see hahttp.sh and the other package control scripts.

Perhaps attach.

SEP

[MEHEE]

The hahttp.sh is the one found in the apache toolkit.

For my configuration:
Two proliant DL380 G4 with a MSA 500
Node 1 : Fep01
eth0 192.168.10.100
eth1 192.168.10.101
eth2 192.168.10.102
eth3 192.168.10.103
bond0 192.168.10.69 (master of eth0 and eth2)
bond1 192.168.0.99 (master of eth1 and eth3)

Node 2 : Fep02
eth0 192.168.10.200
eth1 192.168.10.201
eth2 192.168.10.202
eth3 192.168.10.203
bond0 192.168.10.70 (master of eth0 and eth2)
bond1 192.168.0.98 (master of eth1 and eth3)

My control.sh:
VG[0]="vgpkgA";
LV[0]="/dev/vgpkgA/lvol0"; FS[0]="/mnt/shared"; FS_TYPE[0]="ext3";
FS_MOUNT_OPT[0]="-o rw"; FS_UMOUNT_OPT[0]=""; FS_FSCK_OPT[0]="";
IP[0]="192.168.0.97"
SUBNET[0]="192.168.0.0"
HA_APP_SERVER="post-IP"

My hahttp.conf
SERVER_ROOT="/mnt/shared/httpd"
PID_FILE="/usr/local/cmcluster/conf/pkg2/httpd/httpd.pid"
HTTP_SERVICE_NAME="FAILOVER_HTTP_SERVICE"
HTTP_SERVICE_CMD="/usr/local/cmcluster/conf/pkg2/hahttp.mon"
HTTP_SERVICE_RESTART="-r 0"

With the "same" configuration, I succeeded for the mysql toolkit.

MEHEE

[Steven E. Protter]

Found a doc:

http://www.google.com/url?sa=t&ct=res&cd=2&url=http%3A%2F%2Fdocs.hp.com%2Fen%2FT2342-90001%2FT2342-90001.pdf&ei=9ldsRPLfMc_EwQH-h-neAg&sig2=J6b2wjnaWGXzBsyMUPs1yA

SEP

[MEHEE]

I used the version A.01.00 of this document. With this, i configured the package to use apache and i follow this document step to step.

In this document, i don't find why the package don't start.


MEHEE
PS: I used an identic document for mysql and this package start correctly.

[Steven E. Protter]

This is strange.

Perhaps check for updates on the cluster control toolkit for apache. I just downloaded the thing but have yet to test it.

SEP

[Jim Garrison]

I am a little concerned about your network configuration:

Node 1 : Fep01
eth0 192.168.10.100
eth1 192.168.10.101
eth2 192.168.10.102
eth3 192.168.10.103
bond0 192.168.10.69 (master of eth0 and eth2)
bond1 192.168.0.99 (master of eth1 and eth3)

Node 2 : Fep02
eth0 192.168.10.200
eth1 192.168.10.201
eth2 192.168.10.202
eth3 192.168.10.203
bond0 192.168.10.70 (master of eth0 and eth2)
bond1 192.168.0.98 (master of eth1 and eth3)

Did you really configure your bond interfaces to interfaces that already have IP addresses configured? If so, that probably won't work. My understanding is that the interfaces used in a bond aggregate must not have any IP addresses defined on them.

I am curious if the IP addresses you configured for the bond interfaces are defined in your Serviceguard configuration? If so, then look in your syslogs and see if there are any messages from Serviceguard about those bond interfaces.

I'm thinking your Apache problem might be due to the IP address you're trying to use for Apache not being properly configured.

JimG