Hi, we have a problem recently with Sun Fire T200 server, when installing 10_SPARC_EIS-Jul09 patchbundle resulted for 14 hours of outage! Steps, that was done by me is pretty straightforward:
reboot -- -s
./installcluster --s10cluster
Then I made 1 reboot after kernel patch was installed (it was 14th or 15th to install, but it took server 2 hours to get to it!), then after restarting to single user mode, system took 14 hours to install ~320 patches!!.
There were no errors, logs are clean and identical system with similar 2 zones with WebSphere running was patched for ~2.5 hours...
I had the same problem. I switched to using liveupgrade to patch the servers. Of course this only works if the system has a mirrored root environment. If you can get liveupgrade setup and working, you can patch the inactive boot environment. Then the outage will only be as long as it takes to boot the server.
You can also look into detaching the zones before you install the patches, and then doing a zoneadm attach -u to update the zones after the patches are installed in the global.
Yes Live upgrade is definitely the best practice here. Note that with your root file system on ZFS (Solaris 10 10/08 or later update), you no more need a mirrored boot environment. A ZFS clone will be used instead.
Hi, I must apologize for the degree of hysteria in my first post. I was informed the wrong time, that other system with 3 zones took, it was ~9.5 hours; system I was writing about took ~12 hours with 4 zones, so it all make sense.
I`m very eager to try to upgrade system from this June EIS -> October EIS with parallel patching feature.
If it works as it supposed - it really is a timesaver.