Description of problem: In ovirt3.5 we moved from the use of remoteFileHandlers to use IOProcess for NFS. It was implemented so that we used one instance of IOProcess for all storage domains. We want to change it so each storage domain will have its own instance of IOProcess.
(In reply to Yeela Kaplan from comment #0) > We want to change it so each storage domain will have its own instance of > IOProcess. Why?
This feature will eliminate the dependency between the different storage domains. When different SDs use a single IOProcess, one non-responsive SD will affect all others, as it might take up all available IOProcess threads. Now that each domain has its own IOProcess instance, it will not affect any of the other domains.
Yeela, can you please provide steps for verification for QA or this is more of internal code change? Thanks
When running vdsm with several file SDs you'll be able to see that you have a separate process (IOProcess instance) for each storage domain (+ the global IOProcess instance of course). Vs before when you had a single IOProcess instance for all file SDs.
verified on [root@slot-10 admin]# rpm -q vdsm vdsm-4.17.12-0.el7ev.noarch When new SD attached to engine new IOProcess spawned on host
verified for power, 4 SD +1 Global = 5 IOProcess visible on host vdsm-4.17.19-0.el7ev.noarch ibm-p8-rhevm-FQDN 3.10.0-327.10.1.el7.ppc64le
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2016-0362.html