I am in receipt of 5 new shiney T4-1 Servers, pre-loaded with Solaris 11 which I am trying to configure up.
Hitting an apparent snag when I look to do anything with IPS.
Followed a series of instructions, which are in essence:
1) Download the 2 part ISO files from Oracle.
2) Cat the .iso files together to make a full repository iso.
3) Mount the ISO as hsfs (/mnt)
4) RSync the contents of the mounted ISO to a zfs location (/rpool/repository)
5) Rebuild the repository location with 'pkgrepo rebuild'
6) Check the repository contents with 'pkgrepo info'
root@euripedes:/rpool/repository/repo/publisher/solaris/pkg# pkgrepo info -s /rpool/repository/repo
PUBLISHER PACKAGES STATUS UPDATED
solaris 4401 online 2013-07-04T11:02:08.212013Z
7) Configure the publisher to local file-based:
root@euripedes:/rpool/repository/repo/publisher/solaris/pkg# pkg publisher
PUBLISHER TYPE STATUS P LOCATION
solaris origin online F file:///rpool/repository/repo/
Now, if I try and install ANY package, I get 100 returns of the following error message type:
root@euripedes:/rpool/repository/repo/publisher/solaris/pkg# pkg install motif
Packages to install: 2
Create boot environment: No
Create backup boot environment: No
DOWNLOAD PKGS FILES XFER (MB) SPEED
library/motif 0/2 309/743 0.5/5.0 cache
Errors were encountered while attempting to retrieve package or file data for
the requested operation.
1: file protocol error: code: 2 reason: No file could be found for the specified hash name: 'a9e2b148c65eb17988a80f126eb4f4628c2c5ace'.
Repository URL: 'file:///rpool/repository/repo'. (happened 4 times)
The error lines above get repeated 100 times, with different hash name values.
Whilst I have downloaded the files multiple times, checked the md5 checksums, re-installed the repository from the isos, I continue to get the same errors over and over (even trying other install targets).
Is something getting corrupted, or I am missing something somewhere?
One thing to double check is the permissions on your repository. They should be something like the following:
drwxr-xr-x 2 root root 6 Apr 30 16:19 catalog
drwxr-xr-x 258 root root 258 Mar 5 00:33 file
drwxr-xr-x 2 root root 87 Apr 30 16:23 index
drwxr-xr-x 4423 root root 4423 Apr 14 23:51 pkg
drwxr-xr-x 2 root root 3 Apr 30 16:19 tmp
drwxr-xr-x 2 root root 2 Apr 14 23:51 trans
Is your base system is Oracle Solaris 11 11/11 or Oracle Solaris 11.1? Are you trying to use a the wrong repository version?
The version on the host is currently:
root@euripedes:/rpool/repository# uname -a
SunOS euripedes 5.11 11.1 sun4v sparc sun4v
I did actually wonder whether (being pre-installed) it might have been a different version, but apparently not.
The repository files I have downloaded were from edelivery.oracle.com, under Solaris SPARC, and had part tags of V35100-01 and V35101-01.
I re-downloaded again today, went through all the same process. This time however, I have enabled a host service to serve the repository over http:
root@euripedes:/rpool/ldoms/iso# svcs|grep server
online Jul_03 svc:/network/rpc/smserver:default
online Jul_03 svc:/milestone/multi-user-server:default
online Jul_03 svc:/application/pkg/server:default
root@euripedes:/rpool/ldoms/iso# svcprop -p pkg/inst_root application/pkg/server
PUBLISHER TYPE STATUS P LOCATION
solaris origin online F http://localhost/
Connecting via http to the host shows me the expected webpage with the repository information.
Looking at the permissions in the repository locale; these appear to match what you have listed:
drwxr-xr-x 130 root root 130 Sep 20 2012 file
drwxr-xr-x 4403 root root 4403 Sep 20 2012 pkg
drwxr-xr-x 2 root root 2 Sep 20 2012 trans
drwxr-xr-x 2 root root 6 Jul 5 12:55 catalog
drwxr-xr-x 2 root root 3 Jul 5 12:55 tmp
drwxr-xr-x 2 root root 83 Jul 5 13:00 index
We currently have an issue going through Support. We have had to raise an SR to try and resolve getting our new CSI for the new kit onto our account; so not much help available from that direction just yet. :-o
I have realised the error of my ways: blindly following the documentation.
A number of white paper/pdf documents that I have read through all appear to have had the command syntax:
cat isoa isob > isoc
This appears to have inadvertently caused the overwriting of what was in isoa by what is in isob - so you only actually end up with 1/2 of the repository!!
I finally realised this when I actually took the step to mount each iso (Part A, Part B) individually and rsync'd the contents to the ZFS filesystem. I had noticed that the filesystem space usage was only about the size of the contents of one of the ISOs, and it eventually dawned on me! The resultant space used was then 6.4 GB I think as opposed to 3.6 GB previously.
All appears to be "sorted" now.
Thank you very much for your assistance and kind follow-ups.
Good catch. This has been tripping up a few people. We actually have different download choices in terms of downloading the IPS repository (due to various download restrictions and maximum file systems on some of our hosting servers). There is in fact 2 downloads - one are partial ISO images that can be merged together using cat, the other are individual ISO images that need to be mounted and rsync'd. We've got some proposals for changes coming down the track, so hopefully we can avoid this in future.