I have a big glue workflow (about 100 jobs / crawlers), and it was executing properly until last week. Since then, my first conditional trigger (ALL), is executing 20 time the same job.
I've configured the job it self, to just allow 1 parallel execution, but every time the workflow executes, it tries to launch 20 times (the same job).
Also configured the workflow, to allow a max concurrency of 1, but that doesn't fix the problem.
Since i started working with glue workflow, i've noticed that the tool it self is buggy, old and maybe deprecated ?
Any tips on how to fix this problem ?
I too have faced similar problems. Even one is fixed, some other issue would occur lately. So my suggestion is to try using step functions.
Related
I am deploying a pipeline to Google Cloud DataFlow using Apache Beam. When I want to deploy a change to the pipeline, I drain the running pipeline and redeploy it. I would like to make this faster. It appears from the logs that on each deploy DataFlow builds up new worker nodes from scratch: I see Linux boot messages going by.
Is it possible to drain the pipeline without tearing down the worker nodes so the next deployment can reuse them?
rewriting Inigo's answer here:
Answering the original question, no, there's no way to do that. Updating should be the way to go. I was not aware it was marked as experimental (probably we should change that), but the update approach has not changed in the last 3 i have been using DF. About the special cases of update not working, supposing your feature existed, the workers would still need the new code, so no really much to save, and update should work in most of the other cases.
I have an AWS Glue job, with max concurrent runs set to 1. The job is currently not running. But when I try to run it, I keep getting the error: "Max concurrent runs exceeded".
Deleting and re-creating the job does not help. Also, other jobs in the same account run fine, so it cannot be a problem with account wide service quotas.
Why am I getting this error?
I raised this issue with AWS support, and they confirmed that it is a known bug:
I would like to inform you that this is a known bug, where an internal distributed counter that keeps track of job concurrency goes into a stale state due to an edge case, causing this error. Our internal Service team has to manually reset the counter to fix this issue. Service team has already added the bug fix in their product roadmap and will be working on it. Unfortunately I may not be able to comment on the ETA on the deployment, as we don’t have any visibility on product teams road map and fix release timeline.
The suggested workarounds are:
Increase the max concurrency to 2 or higher
Re-create the job with a different name
Glue container is start and its taking some time same when your job end container shutdown taking some time in between if you try to execute new Jon and default concurrency is 1 so you will get this error.
How to resolve:
Go to your Glue Job --> Under Job detail tab you can find "Maximum concurrency" default value is 1 change it to 3 or more as per your need.
I tried changing "Maximum concurrency" to 2 and then run it !
It worked but again running it cause the same issue, but I looked into my s3 ,it has dumped the data ,so it run for once!
I'm still looking for a stable solution but this may work!
TLDR: I have a pyspark job that finishes in 10 minutes when I run it in a ec2 instance with 16 vcpus but freezes out (it doesn't fail, just never finishes) if I use an instance with over 20 vcpus. I have tried everything I could think of and I just don't know why this happens.
Full story:
I have around 200 small pyspark jobs that for a matter of costs and flexibility I execute using aws batch with spark dockers instead of EMR. Recently I decided to experiment around the best configuration for those jobs and I realized something weird: a job that finished quickly (around 10 minutes) with 16 vcpus or less would just never end with 20 or more (I waited for 3 hours). First thing I thought is that it could be a problem with batch or the way ecs-agents manage the task, so I tried running the docker in an ec2 directly and had the same problem. Then I thought the problem was with the docker image, so I tried creating a new one:
First one used with spark installed as per AWS glue compatible version (https://aws-glue-etl-artifacts.s3.amazonaws.com/glue-1.0/spark-2.4.3-bin-hadoop2.8.tgz)
New one was ubuntu 20 based with spark installed from the apache mirror (https://apache.mirror.digitalpacific.com.au/spark/spark-$SPARK_VERSION/spark-$SPARK_VERSION-bin-hadoop$HADOOP_VERSION.tgz)
Same thing happened. Then I decided the problem was with using docker at all, so I installed everything directly in the ec2, had the same result. Tried changing spark version, also the same thing happened. Thought it could be a problem with hardware blocking too many threads, so I switched to an instance with AMD, nothing changed. Tried modifying some configurations, memory amount used by driver, but it always has the same result: 16 vcpus it work, more than it, it stops.
Other details:
According to the logs it seems to always stop at the same point: a parquet read operation on s3, but the parquet file is super small (> 1mb) so I don't think that is the actual problem.
After that it still has logs sometimes but nothing really useful, just "INFO ContextCleaner: Cleaned accumulator".
I use s3a to read the files from s3.
I don't get any errors or spark logs.
I appreciate any help on the matter!
Stop using the Hadoop 2.7 binaries. They are woefully obsolete, especially for S3 connectivity. replace all the hadoop 2.7 artifacts with Hadoop 2.8 ones, or, preferably, Hadoop 3.2 or later, with the consistent dependencies.
set `spark.hadoop.fs.s3a.experimental.fadvise" to random.
If you still see problems, see if you can replicate them on hadoop 3.3.x, and if so: file a bug.
(advice correct of 2021-03-9; the longer it stays in SO unedited, the less it should be believed)
I have a use case where I schedule a task 24h into the future after an event occurs. This task represents some sort of "deadline" for other things to happen.
The scheduled task triggers a creation of a report. If not all of the above mentioned "other things" have completed by this time, then the triggered report creation process creates it anyways with the information it has at the time.
If, on the other hand, all other things do complete before these 24h, then ideally I'd like to re-use the same Google Cloud Task to trigger the same process (as it's identical as the previous case but will contain all of the information possible).
I would imagine the easiest way to achieve the above is to:
schedule a task 24h into the future
if all information arrives: run the task early before it's scheduled time
However, reading through the Google Cloud Tasks documentation I don't see the option to run the task early. However, that feature does exist on the Cloud Tasks console, so I was wondering if it is available in the documentation and client libraries.
Thanks!
This is probably what you're looking for
https://cloud.google.com/tasks/docs/reference/rest/v2/projects.locations.queues.tasks/run
NOTE: It does say however that "This command is meant to be used for manual debugging"
I have a linear three step Dataflow pipeline - for some reason the last step started, but the preceding two steps hung in Not started for a long time before I gave up and killed the job. I'm not sure what caused this, as this same pipeline had successfully run in the past, and I'm surprised it didn't show any errors in the logs as to what was preventing the first two steps from starting. What can cause such a situation and how can I prevent it from happening?
This was happening because of an error in the worker start up. Certain Dataflow steps do not seem to require workers (e.g. writing to GCS), which is why that step was able to start - i.e. that step starting does not imply that workers are being created correctly. Worker start up is not displayed in the job logs by default - you need to click the link to Stackdriver in the job logs and then add worker-startup in the logs drop down in order to see any of those errors.