UM-Bridge/umbridge

Issues

• HPC: Sporadic model crashes on Helix

  #48 opened 5 months ago by linusseelinger
  0

• HPC: Port selection bug in job.sh

  #83 opened 5 months ago by chun9l
  0

• HPC: Add multiple backend options (e.g. HyperQueue and pure SLURM)

  #80 opened 5 months ago by Schlevidon
  0

• umridge.h: Pass json config by const reference

  #81 opened 8 months ago by Schlevidon
  0

• HPC: Ensure load balancer finds hq when located in same directory

  #59 opened a year ago by linusseelinger
  3

• Python server: Ensure that the model returns a list of lists

  #77 opened 10 months ago by linusseelinger
  0

• Print what job files are found/used

  #70 opened 10 months ago by linusseelinger
  0

• HPC: Enforce FIFO order in jobs

  #65 opened a year ago by linusseelinger
  0

• umbridge.h: Throw proper error when port in serveModels is already occupied

  #68 opened a year ago by linusseelinger
  0

• HPC: Make load balancer available as CI-built statically linked binary

  #52 opened a year ago by linusseelinger
  4

• HPC: Broken/Insufficient Testing of Load Balancer Functionality

  #63 opened a year ago by Schlevidon
  1

• Remove umbridge.h copy in hpc setup

  #53 opened a year ago by linusseelinger
  0

• Avoid file transfers in hpc load balancer

  #26 opened a year ago by linusseelinger
  1

• Remove URL dependency on filesystem for HPC loadbalancer implementation

  #62 opened a year ago by Crambor
  0

• Docs: Document dimensions (and math) of gradient, jacobian action and hessian action

  #61 opened a year ago by linusseelinger
  0

• HPC: Add CI workflow to check building the load balancer works

  #58 opened a year ago by linusseelinger
  0

• Investigate multi-node support in hpc platform

  #30 opened a year ago by linusseelinger
  1

• Skip error checks in hpc load balancer

  #24 opened a year ago by linusseelinger
  1

• Merge hpc load balancer's umbridge.h with main umbridge.h

  #23 opened a year ago by linusseelinger
  0

• Finish hyperuqeue jobs instead of cancelling in load balancer

  #51 opened a year ago by linusseelinger
  0

• Remove HTTP timeout for slow models entirely

  #50 opened a year ago by linusseelinger
  0

• Change timeout for slow job allocations

  #35 opened a year ago by chun9l
  1

• Removing URL files when HQ jobs are cancelled

  #47 opened a year ago by Schlevidon
  1

• Multiple allocation queues create too many SLURM jobs

  #40 opened a year ago by Schlevidon
  2

• HPC: Delete url files at start of load balancer

  #41 opened a year ago by linusseelinger
  1

• Shut down Hyperqueue server automatically after program stops

  #38 opened a year ago by chun9l
  2

• Stop existing hyperqueue instance before launching a new one to always have a clean start

  #43 opened a year ago by linusseelinger
  0

• Refactor waiting for url files to use C++ instead of bash

  #42 opened a year ago by linusseelinger
  0

• Investigate launching of model instances

  #32 opened a year ago by linusseelinger
  1

• Remove direct SLURM call in HPC load balancer

  #25 opened a year ago by linusseelinger
  2

• Model readiness check in hpc load balancer

  #27 opened a year ago by linusseelinger
  2

• Pass model errors through hpc load balancer to client

  #29 opened a year ago by linusseelinger
  0

• Improve servers' error handling

  #28 opened a year ago by linusseelinger
  0

• pymc version

  #13 opened 2 years ago by krosenfeld-IDM
  1

• pymc-client.py fails on window to sample

  #7 opened 2 years ago by krosenfeld-IDM
  6

• Problem running Tsunami model benchmark

  #6 opened 2 years ago by krosenfeld
  30