Patching Sol 10U3 always corrupts....
I have 1 global zone and 11 normal zones.
I have my root and swap mirrored using svm and I have a zfs mirrored on another partition....
Looks like this....
c1d0s0 / -> c2d0s0 / (25GB SVM Mirror)
c1d0s1 swap -> c2d0s1 swap (2GB SVM Mirror)
c1d0s4 zones -> c2d0s4 zones (245GB ZFS Mirror)
I also have a test machine and a VM that I do patch testing and development on. I can patch these system until the cows come home and everything works just fine. It will download, install, and then let me know which ones will be installed during the next reboot, which it does. Then after a 'shutdown -g0 -y -i6', it initiates shutdown, installs the patches, and reboots just fine...
However, on my production box that does have mirrors (both SVC and ZFS) for / and user/zones FS, I get MAJOR problems every time I have a patch that is supposed to install or update the boot-archive.
憇mpatch update?performs fine, downloads, installs what it can, then gives the message on the production system that during the next reboot patches will finish. I initiate the 憇hutdown 杇0 杫 杋6? it installs the patches, sometimes fails or succeeds on some or all of them, then it says 搖pdating archive? only takes a second, reboots and then I get the dreaded 損anic: cannot mount boot archive?error message. I had up until the last time always been able to recover by booting to safe mode, mounting each side of the mirror and doing the 揵ootadm update-archive 朢 /a?on each of the disks.
The last time I did that however, nothing came up and it took many many hours to get the system to be able to boot again. I went WAY over our production outage window and it rolled back some patches.
I do have the Sun online updates contract, ie: I get all the latest patches put out by Sun, but no Sunsolve or phone support.
Can ANYONE help me on the best way to patch the system and keep it up and running? I have tried to boot into single-user, but then it complains that it can抰 install the patches because the zones are running. I can抰 start the zones because I抦 in single-user mode?
I have also tried to boot into full multi-user mode and then initiate 慽nit S? which leaves the zones up and running, patches, but then again, I can抰 reboot without updating the archive?br>
Does anyone know why this occurs? Is it because I have the / mirrored using SVM and the updating archive is not metadisk aware? Should I un-mirror the / before patching? Would a 憁etadetach d10 d12?be sufficient, or do I need to get rebooted to the /dev/dsk/c1d0s0 physical device?
This is SO frustrating as I am using Sun hardware (x2200 M2). I NEVER have this problem under the SPARC platform, and I sorely miss the jumbo 憍_Recommended.zip?cluster we used to be able to download and install. Things 揓ust worked?back then. It seems that since Sun has went to Grub and this smpatch, I have experienced this problem.
This is repeatable on both 10U2 & 10U3.
Thanks for ANY insight you can provide. I have some patches that I need to apply, but I抦 too scared to apply them, and that is not a good thing to be!
Thanks again!
