I'm using custom algorithm running shipped with Docker image on p2 instance with AWS Sagemaker (a bit similar to https://github.com/awslabs/amazon-sagemaker-examples/blob/master/advanced_functionality/scikit_bring_your_own/scikit_bring_your_own.ipynb)
At the end of training process, I try to write down my model to output directory, that is mounted via Sagemaker (like in tutorial), like this:
model_path = "/opt/ml/model"
model.save(os.path.join(model_path, 'model.h5'))
Unluckily, apparently the model gets too big with time and I get the
following error:
RuntimeError: Problems closing file (file write failed: time = Thu Jul
26 00:24:48 2018
00:24:49 , filename = 'model.h5', file descriptor = 22, errno = 28,
error message = 'No space left on device', buf = 0x1a41d7d0, total
write[...]
So all my hours of GPU time are wasted. How can I prevent this from happening again? Does anyone know what is the size limit for model that I store on Sagemaker/mounted directories?
When you train a model with Estimators, it defaults to 30 GB of storage, which may not be enough. You can use the train_volume_size param on the constructor to increase this value. Try with a large-ish number (like 100GB) and see how big your model is. In subsequent jobs, you can tune down the value to something closer to what you actually need.
Storage costs $0.14 per GB-month of provisioned storage. Partial usage is prorated, so giving yourself some extra room is a cheap insurance policy against running out of storage.
In the SageMaker Jupyter notebook, you can check free space on the filesystem(s) by running !df -h. For a specific path, try something like !df -h /opt.
Related
We have a Vertex AI model that takes a relatively long time to return a prediction.
When hitting the model endpoint with one instance, things work fine. But batch jobs of size say 1000 instances end up with around 150 504 errors (upstream request timeout). (We actually need to send batches of 65K but I'm troubleshooting with 1000).
I tried increasing the number of replicas assuming that the # of instances handed to the model would be (1000/# of replicas) but that doesn't seem to be the case.
I then read that the default batch size is 64 and so tried decreasing the batch size to 4 like this from the python code that creates the batch job:
model_parameters = dict(batch_size=4)
def run_batch_prediction_job(vertex_config):
aiplatform.init(
project=vertex_config.vertex_project, location=vertex_config.location
)
model = aiplatform.Model(vertex_config.model_resource_name)
model_params = dict(batch_size=4)
batch_params = dict(
job_display_name=vertex_config.job_display_name,
gcs_source=vertex_config.gcs_source,
gcs_destination_prefix=vertex_config.gcs_destination,
machine_type=vertex_config.machine_type,
accelerator_count=vertex_config.accelerator_count,
accelerator_type=vertex_config.accelerator_type,
starting_replica_count=replica_count,
max_replica_count=replica_count,
sync=vertex_config.sync,
model_parameters=model_params
)
batch_prediction_job = model.batch_predict(**batch_params)
batch_prediction_job.wait()
return batch_prediction_job
I've also tried increasing the machine type to n1-high-cpu-16 and that helped somewhat but I'm not sure I understand how batches are sent to replicas?
Is there another way to decrease the number of instances sent to the model?
Or is there a way to increase the timeout?
Is there log output I can use to help figure this out?
Thanks
Answering your follow up question above.
Is that timeout for a single instance request or a batch request. Also, is it in seconds?
This is a timeout for the batch job creation request.
The timeout is in seconds, according to create_batch_prediction_job() timeout refers to rpc timeout. If we trace the code we will end up here and eventually to gapic where timeout is properly described.
timeout (float): The amount of time in seconds to wait for the RPC
to complete. Note that if ``retry`` is used, this timeout
applies to each individual attempt and the overall time it
takes for this method to complete may be longer. If
unspecified, the the default timeout in the client
configuration is used. If ``None``, then the RPC method will
not time out.
What I could suggest is to stick with whatever is working for your prediction model. If ever adding the timeout will improve your model might as well build on it along with your initial solution where you used a machine with a higher spec. You can also try using a machine with higher memory like the n1-highmem-* family.
I recently started to work with AWS Data Migration Service (DMS) and running into some issues.
Currently attempting to migrate a 10GB Oracle DB to AWS RDS Postgres. Works but has crazy(?) memory requirements. Feels like it loads the entire DB into memory... Started with dms.r4.large (15.5GB) but can not allocate memory after approx. 98%.... Will run smoothly with dms.r4.xlarge (30.5GB)
As you can see in the screenshot (free-able memory, minimum), the instance is constantly running "full" before all memory gets released when the task finishes (or crashs).
Is there any setting to change this and why does it behave like this? It makes the whole task unnecessary expensive...
As confirmed by AWS, this was indeed a bug with the latest engine (v3.1.3). Following additional insights have been provided by AWS to estimate the actual memory requirements:
Full LOB mode (using single row insert+update, commit rate)
Memory: (# of lob columns in a table) x (Number of table in parallel,
default is 8) x (lob chunk size) x (Commit rate during full load) = 2
* 8 *64(k) * 10000k
Note: You may consider to reduce the "Commit rate during full load "
value because we allocate memory using roughly the above method
Limited LOB mode (using array)
Memory: (# of lob columns in a table) x (Number of table in
parallel, default is 8) x maxlobSize x bulkArraySize = 2 * 8 * 4096(k)
* 1000
I want to create Cron in chef witch they verify size of the log if it's > 30mb it will delete it, here is my code:
cron_d 'ganglia_tomcat_thread_max' do
hour '0'
minute '1'
command "rm - f /srv/node/current/app/log/simplesamlphp.log"
only_if { ::File.size('/srv/node/current/app/log/simplesamlphp.log').to_f / 1024000 > 30 }
end
Can you help me in it please
Welcome to Stackoverflow!
I suggest you to go with existing tools like "logrotate". There is a chef cookbook available to manage logrotate.
Please note, that "cron" in chef manages the system cron service which runs independently of chef. You'll have to do the file size check within the "command". It's also better to use the cron_d resource as documented here.
In the way you create cron_d resource it will add cron task only when your log file has size greater than 30mb. In all other cases cron_d will be not created.
You can check that ruby code
File.size('file').to_f / 2**20
to get the file size in Megabytes - there is a slight difference in the result I believe that is more correct.
so you can go wirh 2 solutions for your specific case
create new cron_d resource when log file is less than 30 mb to remove existing cron and provision your node periodically
move the check of the file size in the command with bash and glue with && - in that case file will be dated only if size is greater than 30mb. something like that
du -k file.txt | cut -f1
will return size of the file in bytes
To me also correct way to to that is to use logrotate service and chef recipe for that.
I am running an Aerospike cluster in Google Cloud. Following the recommendation on this post, I updated to the last version (3.11.1.1) and re-created all servers. In fact, this change cause my 5 servers to operate in a much lower CPU load (it was around 75% load before, now it is on 20%, as show in the graph bellow:
Because of this low load, I decided to reduce the cluster size to 4 servers. When I did this, my application started to receive the following error:
All batch queues are full
I found this discussion about the topic, recommending to change the parameters batch-index-threads and batch-max-unused-buffers with the command
asadm -e "asinfo -v 'set-config:context=service;batch-index-threads=NEW_VALUE'"
I tried many combinations of values (batch-index-threads with 2,4,8,16) and none of them solved the problem, and also changing the batch-index-threads param. Nothing solves my problem. I keep receiving the All batch queues are full error.
Here is my aerospace.conf relevant information:
service {
user root
group root
paxos-single-replica-limit 1 # Number of nodes where the replica count is automatically reduced to 1.
paxos-recovery-policy auto-reset-master
pidfile /var/run/aerospike/asd.pid
service-threads 32
transaction-queues 32
transaction-threads-per-queue 4
batch-index-threads 40
proto-fd-max 15000
batch-max-requests 30000
replication-fire-and-forget true
}
I use 300GB SSD disks on these servers.
A quick note which may or may not pertain to you:
A common mistake we have seen in the past is that developers decide to use 'batch get' as a general purpose 'get' for single and multiple record requests. The single record get will perform better for single record requests.
It's possible that you are being constrained by the network between the clients and servers. Reducing from 5 to 4 nodes reduced the aggregate pipe. In addition, removing a node will start cluster migrations which adds additional network load.
I would look at the batch-max-buffer-per-queue config parameter.
Maximum number of 128KB response buffers allowed in each batch index
queue. If all batch index queues are full, new batch requests are
rejected.
In conjunction with raising this value from the default of 255 you will want to also raise the batch-max-unused-buffers to batch-index-threads x batch-max-buffer-per-queue + 1 (at least). If you do not do that new buffers will be created and destroyed constantly, as the amount of free (unused) buffers is smaller than the ones you're using. The moment the batch response is served the system will strive to trim the buffers down to the max unused number. You will see this reflected in the batch_index_created_buffers metric constantly rising.
Be aware that you need to have enough DRAM for this. For example if you raise the batch-max-buffer-per-queue to 320 you will consume
40 (`batch-index-threads`) x 320 (`batch-max-buffer-per-queue`) x 128K = 1600MB
For the sake of performance the batch-max-unused-buffers should be set to 13000 which will have a max memory consumption of 1625MB (1.59GB) per-node.
Referring to the docs, you can specify the number of concurrent connection when pushing large files to Amazon Web Services s3 using the multipart uploader. While it does say the concurrency defaults to 5, it does not specify a maximum, or whether or not the size of each chunk is derived from the total filesize / concurrency.
I trolled the source code and the comment is pretty much the same as the docs:
Set the concurrency level to use when uploading parts. This affects
how many parts are uploaded in parallel. You must use a local file as
your data source when using a concurrency greater than 1
So my functional build looks like this (the vars are defined by the way, this is just condensed for example):
use Aws\Common\Exception\MultipartUploadException;
use Aws\S3\Model\MultipartUpload\UploadBuilder;
$uploader = UploadBuilder::newInstance()
->setClient($client)
->setSource($file)
->setBucket($bucket)
->setKey($file)
->setConcurrency(30)
->setOption('CacheControl', 'max-age=3600')
->build();
Works great except a 200mb file takes 9 minutes to upload... with 30 concurrent connections? Seems suspicious to me, so I upped concurrency to 100 and the upload time was 8.5 minutes. Such a small difference could just be connection and not code.
So my question is whether or not there's a concurrency maximum, what it is, and if you can specify the size of the chunks or if chunk size is automatically calculated. My goal is to try to get a 500mb file to transfer to AWS s3 within 5 minutes, however I have to optimize that if possible.
Looking through the source code, it looks like 10,000 is the maximum concurrent connections. There is no automatic calculations of chunk sizes based on concurrent connections but you could set those yourself if needed for whatever reason.
I set the chunk size to 10 megs, 20 concurrent connections and it seems to work fine. On a real server I got a 100 meg file to transfer in 23 seconds. Much better than the 3 1/2 to 4 minute it was getting in the dev environments. Interesting, but thems the stats, should anyone else come across this same issue.
This is what my builder ended up being:
$uploader = UploadBuilder::newInstance()
->setClient($client)
->setSource($file)
->setBucket($bicket)
->setKey($file)
->setConcurrency(20)
->setMinPartSize(10485760)
->setOption('CacheControl', 'max-age=3600')
->build();
I may need to up that max cache but as of yet this works acceptably. The key was moving the processor code to the server and not relying on the weakness of my dev environments, no matter how powerful the machine is or high class the internet connection is.
We can abort the process during upload and can halt all the operations and abort the upload at any instance. We can set Concurrency and minimum part size.
$uploader = UploadBuilder::newInstance()
->setClient($client)
->setSource('/path/to/large/file.mov')
->setBucket('mybucket')
->setKey('my-object-key')
->setConcurrency(3)
->setMinPartSize(10485760)
->setOption('CacheControl', 'max-age=3600')
->build();
try {
$uploader->upload();
echo "Upload complete.\n";
} catch (MultipartUploadException $e) {
$uploader->abort();
echo "Upload failed.\n";
}