amnovikov
December 13th, 2010, 01:17 PM
Hello everyone!
I`ve got problem during automated upgrade process from Eucalyptus 1.6.2 to 2.0.0 on UEC x64 10.0.4LTS to Ubuntu 10.10 server.
In two words - apt-get and other tools(update-manager) find&download and replace existing packages with new ones. With some packages there are configuration changes - so process ask user what config will be - old or new.
So, the problem.
I`ve replace /etc/eucaliptus/eucalyptus.conf ,cause some new values added, and remove JVM_MEM option "512m". But when started some process that reconfig HSQLDB eucaliptus tables, it faced with java heap error, try to reconnect &reconfig, but finally stuck. In some time, after manually running "dpkg --configure eucalyptus-java-common" and removing and then installing it again, DB reconfig process was also not successed, but it ends his work (not with Ctrl+C key) and update somehow finished.
As it obvious in such condition there is now working cloud HSQLDB, but I also has some previous backups.
My question is - can I somehow use these backups to restore previous 1.6 Cloud configuration (Users,Images, etc) and will I have to reconfig it again? If so - what commands can you give in advice?
Further ---------
Has anyone some ideas where I`m mistaken?
On Ubuntu 10.10 server I temporary change apt repos to LTS and install eucalyptus v.1.6.2. Than recreate or check /var/lib/eucalyptus (especially ./keys ./.ssh and ./db directories ), than change back repos and do upgrade - only eucalyptus packages upgraded. Again error with java heap space - possible walrus db is too large:
root@cloud:~# ls -la /var/lib/eucalyptus/db/ total 40344 -rw-r--r-- 1 eucalyptus eucalyptus 41093528 2010-12-10 21:00 eucalyptus_walrus.script Now, there are such errors as - downgraded eucalyptus 1.6.2 doesnt work correctly and even allow to configure itself (web interface doesnt work, euca_conf --register-cluster .. also). Upgraded finally lost all db values and go with a new one made by own.
What should I do to use data from old HSLDB?
Where should I put options for JVM to not overheaped? (in eucalyptus.conf looks like it doesnt matter)
or another comments, please
I`ve got problem during automated upgrade process from Eucalyptus 1.6.2 to 2.0.0 on UEC x64 10.0.4LTS to Ubuntu 10.10 server.
In two words - apt-get and other tools(update-manager) find&download and replace existing packages with new ones. With some packages there are configuration changes - so process ask user what config will be - old or new.
So, the problem.
I`ve replace /etc/eucaliptus/eucalyptus.conf ,cause some new values added, and remove JVM_MEM option "512m". But when started some process that reconfig HSQLDB eucaliptus tables, it faced with java heap error, try to reconnect &reconfig, but finally stuck. In some time, after manually running "dpkg --configure eucalyptus-java-common" and removing and then installing it again, DB reconfig process was also not successed, but it ends his work (not with Ctrl+C key) and update somehow finished.
As it obvious in such condition there is now working cloud HSQLDB, but I also has some previous backups.
My question is - can I somehow use these backups to restore previous 1.6 Cloud configuration (Users,Images, etc) and will I have to reconfig it again? If so - what commands can you give in advice?
Further ---------
Has anyone some ideas where I`m mistaken?
On Ubuntu 10.10 server I temporary change apt repos to LTS and install eucalyptus v.1.6.2. Than recreate or check /var/lib/eucalyptus (especially ./keys ./.ssh and ./db directories ), than change back repos and do upgrade - only eucalyptus packages upgraded. Again error with java heap space - possible walrus db is too large:
root@cloud:~# ls -la /var/lib/eucalyptus/db/ total 40344 -rw-r--r-- 1 eucalyptus eucalyptus 41093528 2010-12-10 21:00 eucalyptus_walrus.script Now, there are such errors as - downgraded eucalyptus 1.6.2 doesnt work correctly and even allow to configure itself (web interface doesnt work, euca_conf --register-cluster .. also). Upgraded finally lost all db values and go with a new one made by own.
What should I do to use data from old HSLDB?
Where should I put options for JVM to not overheaped? (in eucalyptus.conf looks like it doesnt matter)
or another comments, please