Thank you for your reply. I used glusterfs storage and used three copies for storage. When one of my nodes was down, HostedEngine was suspended for a period of time.I/O errors were reported when I tried to start the HostedEngine virtual machine with virsh resume, and returned to normal when I tried to restart HostedEngine.When I tried to restart HostedEngine again, it was back to normal.

xilazz@126.com
 
From: Gianluca Cecchi
Date: 2019-03-13 22:32
To: xilazz
CC: users
Subject: Re: [ovirt-users] VM has been paused due to a storage I/O error
On Wed, Mar 13, 2019 at 3:56 AM <xilazz@126.com> wrote:
Hi, everyone, there is a VM in the HostedEngine virtual machine that has been paused due to a storage I/O error, but engine manager services are normal, what is the problem?
_______________________________________________


Which kind of storage?
Sometimes when you have block based storage (FC, iSCSI) and you have a disk configured as thin provisioned, it could happen if heavy and/or rapid I/O and LVM resize operations don't satisfy timing needed
If this is your case and you think you need enough sustained I/O in one VM you should configure its related disk as preallocated. 
See this thread about 2 years ago for example:

HIH,
Gianluca