I’m just wondering since it’s been about a year since the arrival of enterprise edition and the removal of updates for the community edition, has it increased the subscription count? I’m asking this because I made the mistake of install the “point” release that was released that was supposed to help fix my vcenter issue and it not only didn’t appear to fix it, I now get random HARD mariadb crashes that I have to delete the tc.log file to fix and I still have to restart opennebula every day to prevent the vcenter connection breaking and then I have to reboot opennebula every ~2 weeks when mariadb crashes. I love the product and would love to support it, but the crashing is really leaving a bad taste in my mouth.
Sorry to hear your are having some issues with your deployment on vCenter. What version of OpenNebula are you currently using?
If it’s a pre-5.12 one, remember that you can get the migrators to the latest CE version (184.108.40.206) by submitting this form: https://opennebula.io/get-migration/
I had updated to 220.127.116.11, and that was a big mistake, it actually made things worse. It didn’t fix the vcenter stability and it breaks mariadb transaction log every 2 weeks. Opennebula use to be awesome, but this policy of witholding updates is now incentivizing me to find an alternative. I understand completely how you want to make money, but there has to be a better way to do it then shutting out your “free” users, sticking them with buggy software with “Hey, sorry it’s buggy, We can fix it for only $4k. Don’t want to pay? Here, heres a tiny update that MAY fix it, or make it worse, but you don’t have a choice, so good luck!”
Hopefully the new version comes soon, since I updated to the “bug fix” version, I now get “Server is not responding” errors in the gui.
I am sorry to hear that you have problems, but maybe just simple software update will not be able to solve your issues if the problem lies with the deployment.
Could you, please, provide some logs and try to debug a little?
The community could maybe be more helpful if it had some relevant information.
Thanks for consideration
I don’t believe the issue lies with the deployment, as this is a “vonecloud” deployment and the underlying vcenter itself is perfectly healthy. I know what’s causing the problem, I have a chron job that runs “one restart” every 8hrs, and after about 12-14 days of that, the maridb will lockup and either restarting all the services or a reboot used to fix that. Since I updated to the .03 release, that didn’t fix the problem and now I get random “server not responding” errors, and when mariadb locks up, a reboot or service restart doesn’t fix it, I have to delete the transaction log to get mariadb to start.
So you’re going to ask, why in the world would you have one restart run every 8hrs? and this brings me to my original post. If I don’t do a one restart, opennebula will stop monitoring vcenter every ~24hrs and marks the server as failed, and then turns off all the VMs. This is a known issue with 5.12 and the .03 update is supposed to fix it. I install the update and no, it didn’t fix the issue. And then I read the “current issues” for 18.104.22.168 and here it is still:
vCenter Monitoring Memory Leak
Process vcenter_monitor.rb has a memory leak problem. Fix will be available in the next stable OpenNebula release. The workaround is to periodically restart the process. This can be done by performing a kill operation on the process, no need to restart the OpenNebula service.
I’m not a linux guy so I just do opennebula restart. That’s why I haven’t reached out to the community, it’s a known issue that non-paying users have to wait to be fixed.
Sorry again to hear that you are having this problem with your deployment. As a general rule, I’d say it’s always a good idea to read in full the Release Notes we prepare for each version before you perform an update.
If you are running a production environment and you feel that you need immediate access to the latest EE patches—the ones we are able to develop thanks to the support of our paying customers, to whom we are very grateful—you might want to consider an OpenNebula Subscription for 5.12, which is also a good way to contribute to the project as a whole.
ok, that was the answer I expected to get.
I understand your frustration but vonecloud (currently) is just an evaluation deployment (like minione) and not intended for production use.
Regarding the issue with leaking memory - that is unfortunate and I am sorry for complications rising from that. The current update policy is as it is but the reason why you started this thread is the crashing mysql/mariadb. That is the bottom line (to me at least).
Mariadb is not OpenNebula product and mariadb is the one having issue after the restarts - maybe there is a bug there… It could also mean that the restart is causing too much writes from OpenNebula and it uses up all the disk space?
Again I understand your frustration but this is probably not the best way to go about it.
Could you pls consider to check the disk space and mariadb’s logs and try to debug a little?
You could start another post or create an issue. It would help you and the project.