With a tenant migration (step 6 in rundeck) to DaaS 9 it could fail with message:
DaasError: ‘Tenant migration failed in TA_MIGRATING phase with error: java.lang.IllegalStateException: Guest process exited unexpectedly: 1’
This message does not give you the right information. In these cases always logon to the new tenant appliance via the service provider, and look into /var/log/desktone/setup. In the setup log look for [Pre-migration Validation], there you will find the exact step what failed.
Once we had a tricky failure at the “Verifying slony failover status”. We checked the /var/log/slony before upgrade, and everything looked fine at that moment as you can see in this screenshot
But in the end it turned out there was replication lag… VMware says it cannot fix this and there are improvements in DaaS 9. But before going to DaaS 9 you have to fix this lag.. How do you find and find this lag?
Finding slony lag:
In our case we used a Nagios check for it (http://www.logix.cz/michal/devel/nagios/check-slony-cluster.pl). This check will trigger you at a broken slony or replication lag.
Or login into psql and run query “select st_lag_time from _slony.sl_status;” on all 3 DaaS databases (edb, fdb and avdb)
if the lag is above 30 minutes you will have to reinitialize.
Fixing slony lag:
In both case you need to reinitialize the databases of this particular tenant! This are the step to reinitialize the slony: https://docs.vmware.com/en/VMware-Horizon-DaaS/services/horizondaas.spmanual800/GUID-D85773AC-0B73-43AB-B52F-48958E9D9E85.html
Once the databases are OK again you can retry the DaaS 9 migration step in rundeck.
