----- Original Message ----- You mean, in the log? Did it 'refuse' to start, or simply failed to start? With what conditions exactly you want this error? What if the engine does manage to start (e.g. because you did some weird tweaking and managed to make it run in 256MB)? What if it later fails, after you add a few tens of hosts/VMs? I know I regularly use a 1GB VM for testing setup, never had problems. Generally, red stuff in setup is reserved to real errors - when something failed. How did you measure that? On my current test VM, I have: # ps aux | grep ovirt-engine ovirt 10840 0.0 0.2 207952 2168 ? Ss 10:08 0:00 /usr/bin/python /usr/share/ovirt-engine/services/ovirt-engine/ovirt-engine.py --redirect-output --systemd=notify start ovirt 10879 0.8 52.4 2247040 534892 ? Sl 10:08 0:46 ovirt-engine -server -XX:+TieredCompilation -Xms1g -Xmx1g -XX:PermSize=256m -XX:MaxPermSize=256m -Djava.net.preferIPv4Stack=true -Dsun.rmi.dgc.client.gcInterval=3600000 -Dsun.rmi.dgc.server.gcInterval=3600000 -Djava.awt.headless=true -Djsse.enableSNIExtension=false -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/var/log/ovirt-engine/dump -Djava.util.logging.manager=org.jboss.logmanager -Dlogging.configuration=file:///var/tmp/ovirt-engine/config/ovirt-engine-logging.properties -Dorg.jboss.resolver.warning=true -Djboss.modules.system.pkgs=org.jboss.byteman -Djboss.modules.write-indexes=false -Djboss.server.default.config=ovirt-engine -Djboss.home.dir=/usr/share/jboss-as -Djboss.server.base.dir=/usr/share/ovirt-engine -Djboss.server.data.dir=/var/lib/ovirt-engine -Djboss.server.log.dir=/var/log/ovirt-engine -Djboss.server.config.dir=/var/tmp/ovirt-engine/config -Djboss.server.temp.dir=/var/tmp/ovirt-engine/tmp -Djboss.controller.temp.dir=/var/tmp/ovirt-engine/tmp -jar /usr/share/jboss-as/jboss-modules.jar -mp /var/tmp/ovirt-engine/modules/00-ovirt-engin -modules:/var/tmp/ovirt-engine/modules/01-jboss-as-modules -jaxpmodule javax.xml.jaxp-provider org.jboss.as.standalone -c ovirt-engine.xml ovirt 10900 0.0 0.2 286084 2604 ? Ss 10:08 0:01 /usr/bin/python /usr/share/ovirt-engine/services/ovirt-websocket-proxy/ovirt-websocket-proxy.py --systemd=notify start root 12200 0.0 0.0 112636 964 pts/10 S+ 11:38 0:00 grep --color=auto ovirt-engine # free total used free shared buffers cached Mem: 1019660 936780 82880 0 5068 259496 -/+ buffers/cache: 672216 347444 Swap: 14749692 164328 14585364 So the engine uses 2.2GB of virtual memory, but only 530MB rss, and total system use is, without caches, 670MB. So it's likely that a 512MB VM will not be enough. Thanks for the report! -- Didi

On Tue, 13 May 2014, Sven Kieske wrote: certainly a fair question ... There is also a statement in that setup script as to needed filesystem space which seems to have been simply 'pulled out of the air', rather than documented / explained Or isolate the recommendations to a flat file which is commented, and sourced by the script, so a person can discern the difference between 'hard' requirements, and simple 'recommendations' for a stated use case -- Russ herrold

I am the reporter of the below bz, and the author of the fix [1]. However, I don't know jboss at all - just searched around, found somewhere the option 'deployment-timeout' (not sure where, probably a quick search can find this) and verified that the fix works. It seems like neither any of the reviewers of the change know enough about jboss to do some more analysis. Can anyone else, that knows jboss, please look at this issue? Specifically: 1. Was anything related to the maximum time for deployments changed in relevant jboss versions? 2. What is affected? What isn't? - fedora 19/20(/21?), el6 - ovirt repos (3.3, 3.4, master) and minimum requirements John: If this is reproducible for you, you might want to try the fix [1] (linked to from the bug). If you do, please report - what exact versions (OS, jboss, ovirt, etc) you used, and (according to the logs) how long did the deployment take. Thanks! [1] http://gerrit.ovirt.org/25895 Best Regards, -- Didi ----- Original Message -----

----- Original Message ----- I searched a bit and can't find a sizing guide etc. for the engine. You can find stuff for the hosts if you search a bit. My current guess is that 1. It's based on anecdotal real-world use 2. It's meant to prevent people from wasting time on not-enough-memory (and disk space) issues etc. 3. In practice, people that use ovirt for more than a minimal setup, will have to have some nice hardware for the hosts, and so dedicating part of that to the engine is not a big issue. Especially with hosted-engine where you do not need a dedicated physical machine. No-one prevents anyone from doing some research and publishing the results, you know - e.g. a table showing "An engine managing X VMs on Y hosts used such-and-such disk space over the first day/week/month/year of use, and had this-and-that average response time (or something more complex) when running with such-and-such RAM". If, based on that, you think we can/should provide more info regarding minimal/recommended RAM/disk for specific use cases, you are then welcome to update the wiki. Patches to setup are welcome too :-) Note that current limitations are never failing setup - they are always just warnings. You are welcome to ignore them (and feed that to your answer files if you run setup repeatedly). Not sure I completely got you here, but if you meant to a file summarizing the results of such a hypothetical research, then I think that would be great. Best, -- Didi