Enlarge root volume to allow NetApp SMOS installation

[RobertCarback]

We have purchased NetApp's SMOS (Snap Mirror for Open Systems -- replication) product (formerly named ReplicatorX). We have, first, a Dev server (call it dev1 -- rx6600, HP-UX 11.23 IA64) that has a root volume ("/") sized at 576MB. In order to install SMOS we need an additional 1 GB of available disk on the root volume. I understand that ignite is one way to increase root by doing a make_net_recovery, booting up dev1 from the image on the ignite server, increasing the size of root and restoring vg00, but does anyone have an alternative method to increase root (reinstall is not an option, as this will also have to be done on the 2 nodes in our production Serviceguard cluster)

[Torsten.]

Why you need space in / (root)?

Where will the software be installed?

I don't know this software, but I would guess it will go to /opt or wherever, but not directly to /.

[RobertCarback]

SMOS requires 1 GB of disk space on the root volume.

[Sagar Sirdesai]

Hi Robert

I had smilar problem I had to install JavaJDK and / was full

So I created new mount point /opt/Java42 and installed it using the below command swinstall in /opt/java42

swinstall -s /opt/java.depot JavaJDK,l=/opt/java42.

Where JavaJDK is the bundle name

Please try on this lines

Sagar

[Bill Hassell · Accepted solution]

Rather than look at general statements about "SMOS requires 1GB of root disk", ask to see a list of the fullpath names of all the files. Many vendors do not understand SysV filesystem layouts and that most cpommerical versions of Unix have separate mountpoints for /var, /opt, /tmp and /usr. If indeed the application requires 1 GB in the root volume, I would stop immediately and ask why. This would be a severe impact to virtually any Unix system, requiring a complete re-installation (such as Ignite for HP-UX) and represents a very high risk to your clustered environment. Ask why the installation does not ask the system administrator where the application will be installed.

[RobertCarback]

this is from the vendor (remember this is about installing replication software)---

the rational is .. on Unix you can have dozens and dozens of volumes.. if the server ever crashes or reboots the first thing up is the rootvg.. we want the â kernelâ filter ; this is what â interceptsâ / copies the write on the source and makes sure we keep track of what was written..


so in a Large unix env. you can have a crash.. the rootvg comes up .. which then brings up the .. say the OracleDB.. and this application then does it job and starts writing to its target volume.. what if the volume SMOS is loaded on, is still not mounted ? .. we can have a situation where there were writes on the source .. we

did not replicate to the target volume.

[RobertCarback]

rewritten for clarity --

this is from the vendor (remember this is about installing replication software)---

the rational is .. on Unix you can have dozens and dozens of volumes.. if the server ever crashes or reboots the first thing up is the rootvg.. we want the kernel filter ; this is what intercepts / copies the write on the source and makes sure we keep track of what was written..


so in a Large unix env. you can have a crash.. the rootvg comes up .. which then brings up the .. say the OracleDB.. and this application then does it job and starts writing to its target volume.. what if the volume SMOS is loaded on, is still not mounted ? .. we can have a situation where there were writes on the source .. we

did not replicate to the target volume.

[RobertCarback]

closed. the requirement was for the root VG, not the root lvol