Batch - Issues mounting new disk

Google Cloud Compute Infrastructure
,
1

I am trying to create a job through python, this is my creation script:

def submit_download_sra_fastq(sra_batch_file, batch_name, project_id, region, job_name, docker_image, cpus, memory_mib,
                              machine_type, max_retry_count, run_minutes, parallelism):

    sra_count = 0
    with open(sra_batch_file) as infile:
        for _ in infile:
            sra_count += 1

    if batch_name is None:
        gcs_batch_file = 'gs://sra-fastq/batches/sra-batch'+str(time.time()).split('.')[0]+'.txt'
    else:
        gcs_batch_file = 'gs://sra-fastq/batches/' + batch_name + '.txt'

    command = f'gsutil cp {sra_batch_file} {gcs_batch_file}'
    subprocess.run(command.split())

    client = batch_v1.BatchServiceClient()

    # Define what will be done as part of the job.
    runnable = batch_v1.Runnable()
    runnable.container = batch_v1.Runnable.Container()
    runnable.container.image_uri = docker_image
    runnable.container.options = '--mount type=bind,src=/mnt/disks/sra-download/,target=/sra-download'
    runnable.container.entrypoint = "/bin/sh"

    runnable.container.commands = ["-c", f"hsubatch download-sra-fastq {gcs_batch_file} $BATCH_TASK_INDEX"]

    # Jobs can be divided into tasks. In this case, we have only one task.
    task = batch_v1.TaskSpec()
    task.runnables = [runnable]
    volume = batch_v1.Volume()
    volume.device_name = 'sra-download'
    volume.mount_path = "/mnt/disks/sra-download/"
    volume.mount_options = "rw,async"
    task.volumes = [volume]

    # We can specify what resources are requested by each task.
    resources = batch_v1.ComputeResource()
    resources.cpu_milli = cpus * 1000  # in milliseconds per cpu-second. This means the task requires 2 whole CPUs.
    resources.memory_mib = memory_mib  # in MiB
    task.compute_resource = resources

    task.max_retry_count = max_retry_count
    task.max_run_duration = str(run_minutes * 60) + "s"

    # Tasks are grouped inside a job using TaskGroups.
    # Currently, it's possible to have only one task group.
    group = batch_v1.TaskGroup()
    group.task_spec = task

    group.task_count = sra_count
    group.parallelism = parallelism

    # Policies are used to define on what kind of virtual machines the tasks will run on.
    # In this case, we tell the system to use "e2-standard-4" machine type.
    # Read more about machine types here: https://cloud.google.com/compute/docs/machine-types
    policy = batch_v1.AllocationPolicy.InstancePolicy()
    policy.machine_type = machine_type
    policy.provisioning_model = "SPOT"

    newdisk = batch_v1.AllocationPolicy.AttachedDisk()
    newdisk.new_disk.size_gb = 100
    newdisk.new_disk.type_ = 'pd-ssd'
    newdisk.device_name = 'sra-download'
    policy.disks = [newdisk]

    instances = batch_v1.AllocationPolicy.InstancePolicyOrTemplate()
    instances.policy = policy
    allocation_policy = batch_v1.AllocationPolicy()
    allocation_policy.instances = [instances]

    job = batch_v1.Job()
    job.task_groups = [group]
    job.allocation_policy = allocation_policy
    job.labels = {"env": "testing", "type": "container"}
    # We use Cloud Logging as it's an out of the box available option
    job.logs_policy = batch_v1.LogsPolicy()
    job.logs_policy.destination = batch_v1.LogsPolicy.Destination.CLOUD_LOGGING

    create_request = batch_v1.CreateJobRequest()
    create_request.job = job
    create_request.job_id = job_name
    # The job's parent is the region in which the job will run
    create_request.parent = f"projects/{project_id}/locations/{region}"

    client.create_job(create_request)

    print(f"Created job: {job_name}")

You can see that I am trying to mount a new disk to this instance, and then run a docker image that bind mounts to that new disk. I get these errors when I run the command:

ERROR 2023-01-12T07:21:24.898106623Z mke2fs 1.46.5 (30-Dec-2021)
INFO 2023-01-12T07:21:24.901911705Z Discarding device blocks: 0/26214400
INFO 2023-01-12T07:21:24.966901031Z done
INFO 2023-01-12T07:21:24.967126864Z Creating filesystem with 26214400 4k blocks and 6553600 inodes Filesystem UUID: a26b88d9-e73b-4a34-ba38-5b11c9b6f996 Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912
INFO 2023-01-12T07:21:24.967145137Z , 819200, 884736, 1605632, 2654208, 4096000
INFO 2023-01-12T07:21:24.967153330Z , 7962624,
INFO 2023-01-12T07:21:24.967160683Z 11239424
INFO 2023-01-12T07:21:24.967173601Z , 20480000
INFO 2023-01-12T07:21:24.967189740Z , 23887872 Allocating group tables: 0/800
INFO 2023-01-12T07:21:24.968106547Z done
INFO 2023-01-12T07:21:24.968778284Z Writing inode tables: 0/800
INFO 2023-01-12T07:21:25.002198029Z 68/800
INFO 2023-01-12T07:21:25.174615487Z f927477c601c: Pull complete
INFO 2023-01-12T07:21:25.245055966Z 7eb0e5940f70: Pull complete
INFO 2023-01-12T07:21:25.257080266Z Digest: sha256:3be61617119fbccfd39a8b892debe155abb6e2d96ba7dcd1f1f911b24847a619
INFO 2023-01-12T07:21:25.262943714Z Status: Downloaded newer image for gcr.io/mdurrant/hsulab-gcp-batch:latest
INFO 2023-01-12T07:21:25.266026529Z gcr.io/mdurrant/hsulab-gcp-batch:latest
ERROR 2023-01-12T07:21:25.278811140Z mke2fs 1.46.5 (30-Dec-2021)
INFO 2023-01-12T07:21:25.282597747Z Discarding device blocks: 0/26214400
INFO 2023-01-12T07:21:25.342568710Z done
INFO 2023-01-12T07:21:25.342952775Z Creating filesystem with 26214400 4k blocks and 6553600 inodes Filesystem UUID: 572c3a05-08af-4efb-9e14-24833c696273 Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736,
INFO 2023-01-12T07:21:25.342978304Z 1605632, 2654208, 4096000, 7962624, 11239424, 20480000, 23887872
INFO 2023-01-12T07:21:25.342985559Z Allocating group tables:
INFO 2023-01-12T07:21:25.342992886Z 0/800
INFO 2023-01-12T07:21:25.343864852Z done
INFO 2023-01-12T07:21:25.344516947Z Writing inode tables: 0/800
INFO 2023-01-12T07:21:25.377867873Z done
INFO 2023-01-12T07:21:25.382698785Z Creating journal (131072 blocks):
INFO 2023-01-12T07:21:25.384877891Z done Writing superblocks and filesystem accounting information:
INFO 2023-01-12T07:21:25.385367312Z 0/800
INFO 2023-01-12T07:21:25.387779978Z 
INFO 2023-01-12T07:21:25.395778084Z done
ERROR 2023-01-12T07:21:25.497736569Z mount: /mnt/disks/sra-download: wrong fs type, bad option, bad superblock on /dev/sdb, missing codepage or helper program, or other error.

Looks like there is an issue with mounting the new disk, any idea what is going wrong?

2

Hi,
don’t know Python, but it looks like there is something wrong with FS type itself. Are you defining FS type somewhere under the code ? If not, perhaps there is default type defined which is not supported ?

Maybe here you will be able to define for example ext4 FS type as mke2fs supports only ext2/ext3/ext4 fs types.

best,
DamianS

3

According to https://cloud.google.com/python/docs/reference/batch/latest/google.cloud.batch_v1alpha.types.Volume
and https://man7.org/linux/man-pages/man8/mount.8.html

you should be able to define fs type, so most probably it should work like that :
volume.mount_options = “rw,async,ext4”

4

Thanks for the speedy response. I made the change and got the same error:

mount: /mnt/disks/sra-download: wrong fs type, bad option, bad superblock on /dev/sdb, missing codepage or helper program, or other error.
5

Hi,

Hmm,

Maybe, if you are doing bind via docker , this part with volume device is not needed? Because it looks like code is creating FS twice, first with UUID and blocks:

INFO 2023-01-12T07:21:24.967126864Z Creating filesystem with 26214400 4k blocks and 6553600 inodes Filesystem UUID: a26b88d9-e73b-4a34-ba38-5b11c9b6f996 Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912

and second time with totally different blocks:

NFO 2023-01-12T07:21:25.342952775Z Creating filesystem with 26214400 4k blocks and 6553600 inodes Filesystem UUID: 572c3a05-08af-4efb-9e14-24833c696273 Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736,
INFO 2023-01-12T07:21:25.342978304Z 1605632, 2654208, 4096000, 7962624, 11239424, 20480000, 23887872
6

That would be strange if docker mounted the disk locally and mounted it to the docker container. I don’t think that’s it. I tried using the runnable.container.volumes option in python to bind mount to the docker container and that still did not work.

7

Hi, mgdurrant,

Thanks for the detailed job info! When there is no container volume setting, we set volume from host VM as its default volume. So in this case, after we mount the volume for the container, the mount commands from runnable.container.volumes option will fail.

Could you please try our default mount without specifying python mount OR configuring the volume for container explicitly to avoid the default behavior?

Please let us know if it works for you.

Thanks!

Wen

8

I need more disk space when using the docker container, and I don’t know how to increase host VM boot disk space. That’s why I was trying to mount it to the external disk. Can you advise on that?

9

Hi mgdurrant,

Having a larger size of boot disk can be achieved via an instance template or defining a large boot disk size with boot_disk_mib in compute resources.

Plz let us know if it works. Thanks!

Wen

1 Like
10

Exactly what I was looking for, thank you!