64GB内存服务器无法创建大于m1.tiny规格实例求助（OpenStack Ocata）

Hey there, let's break down the issue you're facing—being able to spin up m1.tiny instances but failing with larger flavors on your 64GB-memory Ubuntu 16.04 + OpenStack Ocata setup. That error about the volume timing out after 0 seconds is a big clue, so let's start there.

Key Observations from the Error #

"Volume 71bc03cf-6ab1-4511-8763-43647fd1ea2c did not finish being created even after we waited 0 seconds or 1 attempts."

The "waited 0 seconds" part is suspicious—it suggests Nova isn't giving Cinder any time to actually create the volume. Let's start with the most likely fix for that.

Step 1: Fix Nova's Volume Creation Timeout Setting #

This is probably the root cause. Nova has a configuration parameter that controls how long it waits for Cinder to create a volume. If it's set to 0, Nova will immediately mark the volume creation as failed.

• Edit your Nova configuration file at /etc/nova/nova.conf
• Look for the [DEFAULT] section and find the volume_create_timeout parameter. If it's set to 0, change it to a reasonable value like 60 (seconds):

  [DEFAULT]
  volume_create_timeout = 60
  

• Restart all Nova services to apply the change:

  sudo systemctl restart openstack-nova-api openstack-nova-compute openstack-nova-scheduler
  

Step 2: Verify Cinder Service Health #

Since larger flavors likely rely on Cinder for root volumes (while m1.tiny might use local ephemeral storage), we need to ensure Cinder is working properly:

• Check if all Cinder services are active and running:

  openstack volume service list
  

  Make sure all entries show up in the Status column.
• Inspect Cinder's volume logs for the problematic volume ID (71bc03cf-6ab1-4511-8763-43647fd1ea2c):

  grep "71bc03cf-6ab1-4511-8763-43647fd1ea2c" /var/log/cinder/volume.log
  

  Look for errors like insufficient storage space, permission issues, or backend driver failures (e.g., if you're using LVM, check if the volume group has free space with vgs).

Step 3: Check Flavor and Image Compatibility #

Larger flavors might require bigger root volumes than m1.tiny. Let's confirm the disk requirements match your image:

• Check the disk size of the flavor you're trying to use:

  openstack flavor show <your-large-flavor-name>
  

  Look at the disk field (this is the required root volume size in GB).
• Compare it to the size of your boot image:

  openstack image show <your-boot-image-name>
  

  If the image size is larger than the flavor's disk allocation, this could cause volume creation failures. Either resize the flavor's disk or use a smaller image.

Step 4: Review Local.conf Deployment Settings #

Since you mentioned your local.conf file, double-check these key configurations:

• If using LVM for Cinder, ensure LVM_VOLUME_GROUP points to a valid volume group with enough free space.
• Confirm the Cinder volume driver is correctly set (e.g., CINDER_VOLUME_DRIVER=cinder.volume.drivers.lvm.LVMVolumeDriver for LVM).
• Verify that Nova is configured to use Cinder for root volumes when needed (check for default_volume_type in nova.conf).

Final Notes #

If the timeout fix doesn't resolve the issue, the logs from Nova (/var/log/nova/nova-compute.log and /var/log/nova/nova-scheduler.log) will give more details about why the instance creation is failing. Pay attention to any errors related to volume attachment or flavor validation.

内容的提问来源于stack exchange，提问作者Dew