PDA

View Full Version : unable to migrate to smt repositories instanceInfraUpgrade



Hikerguy88
18-May-2017, 21:26
we have 3 primary SAP HANA servers running SUSE as EC2 instances on Amazon AWS. I am new to my company and have inherited these servers.

we have a server for dev, test, and production, all are running SLES 11 SP2

I need to install packages, upgrade them, etc. I noticed we were still using susecloud service in zypper


-- Dev Environment --
I did the upgrade procedure but received errors so I did the following:

cd /etc/zypp
mkdir old
mv credentials.d repos.d services.d old
mkdir credentials.d repos.d services.d
instanceInfraUpgrade

(forums.aws.amazon.com had that idea)

I received more errors about missing python-mcrypt package so I downloaded and installed that (wget http://download.opensuse.org/repositories/openSUSE:/Tools/SLE_11_SP2/x86_64/python-M2Crypto-0.21.1-1.1.x86_64.rpm)

installed and ran the instanceUpgradeInfra script again and it worked perfectly. new repos are enabled and work on DEV

zypper returned the correct repos and I could install software again -- great.

-- Test Environment --

downloaded the instanceInfraUpgrade package
when I ran the script it deleted all of the old susecloud repos, but didn't install any new ones.

zypper lr
# | Alias | Name | Enabled | Refresh
--+-------------------------+-------------------------+---------+--------
1 | nVidia-Driver-SLE11-SP2 | nVidia-Driver-SLE11-SP2 | Yes | Yes
2 | server_monitoring | server_monitoring | Yes | No


I can see in the script this code checks for "susecloud" and if it is not found, it assumes (wrongly) that smt repos are installed, so it removes itself and quits



if [ -f /usr/sbin/registercloudguest ]; then
hasSUSEcloud=$(zypper ls | grep -v SMT | grep 'susecloud')
if [ -n "$hasSUSEcloud" ];then
removeDeprecatedService
rm -rf /etc/zypp/repos/*susecloud*
else
echo "--> Instance is already set up for the new infrastructure, nothing todo"
fi
destruct
exit 0
fi


comment that out.

now, the setup function will fail because it has already removed susecloud so it can't remove it again, so we will comment that out along with the check of what that returns:



#suse_register --no-refresh -E || return
...
#if ! setup; then
# echo "--> Setup cloud configuration failed"
# exit 1
#fi


ok now it goes directly to the attempt to start the guestregister service, which fails and creates /var/log/cloudregister

errors are "server error 403 unauthorized" for every IP. It's a 242 line file with errors, I assume I shouldn't post that hahaha.

I also noticed the /etc/zypp/credentials.d/ directory is empy now -- should that be restored from the archive I made of /etc/zypp/* to resolve the errors?



how do I resolve this and get the smt repositories? I need to do this for this environment and production still. Please let me know if anyone has any ideas.

Hikerguy88
19-May-2017, 16:24
Maybe important to note, the test server does not have a public IP (EIP or static given at instance launch) where as the Dev server does have a public IP.

also the error is 403 forbidden

2017-05-19 10:34:44,296 INFO:Using region server: 104.45.154.114
2017-05-19 10:34:44,323 INFO:Using region server: 104.45.31.195
2017-05-19 10:34:44,694 INFO:Using region server: 191.237.254.253
2017-05-19 10:34:45,202 INFO:Using region server: 23.100.36.229
2017-05-19 10:34:45,487 INFO:Using region server: 23.101.26.184

2017-05-19 10:34:45,488 INFO:Starting new HTTPS connection (1): 23.101.26.184
2017-05-19 10:34:46,402 ERROR:====================
2017-05-19 10:34:46,402 ERROR:Server returned: 403

the other server registered to

54.225.105.144 smt-ec2.susecloud.net smt-ec2