[Linux-cluster] Indenting resources inside a service in rh el 4.8

[Gianluca Cecchi]

Probably I was not completely correct in my assumptions.

The indenting configuration should assure that the inner resource is
started after the outer one. Is this true for rh el 4?
Instead, to have the behaviour regarding resource restatrt and not
whole service restart, in rh el 5.x you should write something like
this:

                        <fs ref="extstoragefs" __independent_subtree="1">
                                <script
ref="myscript"__independent_subtree="1" />
                        </fs>
                                <ip ref="10.4.5.6"/>
                                <script ref="mydb"/>

But this seems not to work in rh el 4. Is this supposed to work?
Thanks
Gainluca


On Tue, Oct 12, 2010 at 3:40 PM, Gianluca Cecchi
<gianluca.cecchi at gmail.com> wrote:
> Hello,
> in RH EL 5.4 (and 5.5) it is possible to indent resources inside a
> service definition (service = test_service), so that, having this:
>
>                        <fs ref="extstoragefs">
>                                <script ref="myscript"/>
>                        </fs>
>                                <ip ref="10.4.5.6"/>
>                                <script ref="mydb"/>
>
>
> Where resources are like these ones:
>                <resources>
>                        <fs device="/dev/md0" force_fsck="0"
> force_unmount="1" fsid="12263" fstype="ext3" mountpoint="/app"
> name="extstoragefs" options="" self_fence="1"/>
>                        <ip address="10.4.5.6" monitor_link="0"/>
>                        <script file="/opt/myapp/init.d/myscript.sh"
> name="myscript"/>
>                        <script file="/etc/init.d/mysql" name="mydb"/>
>                </resources>
>
> Suppose I have a failure in myscript app (I kill the app) I have a
> restart in place of the resource itslef.
> Instead it seems that in latest RHEL 4.8+updates I have a complet
> service restart.
> Is this the supposed behaviour so that there is different
> functionality between RH EL 4.x and 5.x?
> I suppose so, as I don't see any particular xml formatting
> informations in cluster admin manual for rh el 4.x.... but I would
> like to have a confirmation.
> Any way to have similar parent-childs configuration in RH EL 4?
>
> In logs, after kiil -9 of the process involved in myscript application I have:
>
> Oct 12 15:07:14 nodeb clurgmgrd: [27555]: <err> script:myscript:
> status of /opt/myapp/init.d/myscript.sh failed (returned 3)
> Oct 12 15:07:14 nodeb clurgmgrd[27555]: <notice> status on script
> "myscript" returned 1 (generic error)
> Oct 12 15:07:14 nodeb clurgmgrd[27555]: <notice> Stopping service test_service
> Oct 12 15:07:16 nodeb lsb_log_message:  succeeded
> Oct 12 15:07:26 nodeb clurgmgrd[27555]: <notice> Service test_service
> is recovering
> Oct 12 15:07:26 nodeb clurgmgrd[27555]: <notice> Recovering failed
> service test_service
> Oct 12 15:07:26 nodeb kernel: kjournald starting.  Commit interval 5 seconds
> Oct 12 15:07:26 nodeb kernel: EXT3-fs warning: maximal mount count
> reached, running e2fsck is recommended
> Oct 12 15:07:26 nodeb kernel: EXT3 FS on md0, internal journal
> Oct 12 15:07:26 nodeb kernel: EXT3-fs: mounted filesystem with ordered
> data mode.
> Oct 12 15:07:31 nodeb lsb_log_message:  succeeded
> Oct 12 15:07:31 nodeb clurgmgrd[27555]: <notice> Service test_service started
>
> Thanks in advance, Gianluca
>