On Thu, Mar 21, 2024 at 12:44 PM Claus Serbe via Users <users@ovirt.org> wrote:
Hi,

I am migrating some vmware VM's from an NFS Storage via rhv-upload in virt-v2v, what is working good.

But now I try to move some bigger VM's with several disks and sadly after a while (I would guess around an hour) the Ovirt-engine shows me "Paused by system" instead of transfering, so when the next disk should be imported,  it will fail

In the ovirt-engine.log I see the following lines for the remaining 4 disks.

2024-03-21 06:14:06,815-04 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-35) [f61b3906-804d-470f-8524-6507081fbdec] EVENT_ID: UPLOAD_IMAGE_PAUSED_BY_SYSTEM_TIMEOUT(1,071), Upload was paused by system. Reason: timeout due to transfer inactivity.
2024-03-21 06:14:17,915-04 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-14) [aef8e312-d811-4a39-b5fb-342157209bce] EVENT_ID: UPLOAD_IMAGE_PAUSED_BY_SYSTEM_TIMEOUT(1,071), Upload was paused by system. Reason: timeout due to transfer inactivity.
2024-03-21 06:14:24,959-04 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-85) [860b012d-78a4-49f8-a875-52f4299c8298] EVENT_ID: UPLOAD_IMAGE_PAUSED_BY_SYSTEM_TIMEOUT(1,071), Upload was paused by system. Reason: timeout due to transfer inactivity.
2024-03-21 06:14:46,099-04 ERROR [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (EE-ManagedScheduledExecutorService-engineScheduledThreadPool-Thread-65) [f93869ee-2ecb-4f54-b3e9-b12259637b0b] EVENT_ID: UPLOAD_IMAGE_PAUSED_BY_SYSTEM_TIMEOUT(1,071), Upload was paused by system. Reason: timeout due to transfer inactivity.


There are 2 strange things.

1. When I start virt-v2v it will create all 6 disks and set them to transferring, but virt-v2v will import one after the other, what leads to kind of unused/timing out transferring tickets.

This is a incorrect usage of ovirt-imageio API in virt-v2v, please report it here:
https://github.com/libguestfs/virt-v2v/issues

The right way to use the API is:
1. Start transfer
2. Upload the data
3. End the transfer

It does not matter if you create all the disk at the start of the operation, but starting a transfer must be done
right before you upload the data. 
 
2. When I copy the disk images to a local disk before, it works. Maybe just because of faster transfer speeds.

Is there a possibility to transfer parallel or maybe extend the timeout?

Sure you can upload in parallel, but I'm not sure virt-v2v will probably have issues importing in parallel from other
systems (e.g. vmware would not allow this).

We tested uploading 10 100g images in parallel using the ovirt-img tool, each upload using 4 connections
(total of 40 upload connections).

You may be able to extend the timeout, but this is not recommended since your system will not clean up quickly
after a bad client disconnects uncleanly without ending the transfer. Unfortunately I don't remember the which 
timeout should be modified on the engine side, maybe Arik or Albert can help with this.

Nir