Respectable Gr-R, thank you for the answer.
Excuse me, but your recommendations is applicable for the test lab, not for production. You position your product as "Enterprise-Class Storage Software for Every Business".
What means "may or may not work" in production environment with many virtual machines hosted on the storage?
What means "stop the cluster services" in production?
It means "Denial of Service" for all servicies with unpredictable results at unpredictable times.
My cluster config is fully consistent with the manufacturer's recommendations.
Failure of one of the nodes in a two-node cluster is fully supported event.
I think, that any enterprise solution must have stadard procedure for returning failed node in the cluster.
My new node hardware configuration is absolutelly identic to the old hardware configuration.
My new node software version is absolutelly identic to the old software version.
I described the current configuration of each of the nodes.
Now, as a customer, i want to recieve a detailed correct instructions how to return new node in the cluster according to my config, not according to any other situation.
Instead of this, you and your support team gives me a vague instructions like "manually create what is missing".
In my post i described in detail what i have.
I just ask to specify what I am missing?
What is the error "Too many bound hosts on remote host."?
What is the error "Connection to host 'dss1-host' lost", although all pings in all directions for all ip addresses is successful and Host binding from "dss2" to "dss1" is reacheable?
I would gladly to send a detailed diagram of my configuration and any logs, but do not have the rights to post attachments. Ready to explain everything that is not clear, in words.
Now will review your basic recommendation to restore config from saved state. It is consistent with production logical configuration. I was not doing any changes in configuration after saving.
BUT, as I wrote above, I fear replication in the wrong direction. Today all source volumes are hosted on "dss2" node as a sources. After restoring config from saved state, lv0000 and lv0001 will be hosted on "dss1" node as a sources. But on "dss1" node they are blank volumes. If, for some reason, replication goes in the wrong direction, I will lose all data on the lv0000 and lv0001.
How safe is recovery from saved state in this situation???
In fact, I have already collected manually configure identical to settings that were saved, except for setting replication origins. Why do you think, that restoring is better than my manual config?
Reply With Quote