dag2_disk_usage.py 2.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106
  1. from datetime import timedelta
  2. from airflow.models import DAG,Variable
  3. from airflow.utils.dates import days_ago
  4. from airflow.operators.bash_operator import BashOperator
  5. from airflow.contrib.operators.ssh_operator import SSHOperator
  6. from airflow.contrib.hooks.ssh_hook import SSHHook
  7. default_args = {
  8. 'owner': 'airflow',
  9. 'depends_on_past': False,
  10. 'start_date': days_ago(2),
  11. 'email_on_failure': False,
  12. 'email_on_retry': False,
  13. 'retries': 1,
  14. 'retry_delay': timedelta(minutes=5)
  15. }
  16. orwell_ssh_hook = \
  17. SSHHook(
  18. key_file=Variable.get('hpc_ssh_key_file'),
  19. username=Variable.get('hpc_user'),
  20. remote_host='orwell.hh.med.ic.ac.uk')
  21. woolf_ssh_hook = \
  22. SSHHook(
  23. key_file=Variable.get('hpc_ssh_key_file'),
  24. username=Variable.get('hpc_user'),
  25. remote_host='woolf.med.ic.ac.uk')
  26. eliot_ssh_hook = \
  27. SSHHook(
  28. key_file=Variable.get('hpc_ssh_key_file'),
  29. username=Variable.get('hpc_user'),
  30. remote_host='eliot.med.ic.ac.uk')
  31. igf_lims_ssh_hook = \
  32. SSHHook(
  33. key_file=Variable.get('hpc_ssh_key_file'),
  34. username=Variable.get('hpc_user'),
  35. remote_host='igf-lims.cc.ic.ac.uk')
  36. dag = \
  37. DAG(
  38. dag_id='dag2_disk_usage',
  39. schedule_interval=None,
  40. max_active_runs=1,
  41. default_args=default_args)
  42. with dag:
  43. check_orwell_disk = \
  44. SSHOperator(
  45. task_id = 'check_orwell_disk',
  46. dag = dag,
  47. ssh_hook = orwell_ssh_hook,
  48. queue='hpc_4G',
  49. command = 'bash /home/igf/igf_code/IGF-cron-scripts/orwell/orwell_disk_usage.sh '
  50. )
  51. check_eliot_disk = \
  52. SSHOperator(
  53. task_id = 'check_eliot_disk',
  54. dag = dag,
  55. ssh_hook = eliot_ssh_hook,
  56. queue='hpc_4G',
  57. command = 'bash /home/igf/git_repos/IGF-cron-scripts/eliot/eliot_disk_usage.sh '
  58. )
  59. check_woolf_disk = \
  60. SSHOperator(
  61. task_id = 'check_woolf_disk',
  62. dag = dag,
  63. ssh_hook = woolf_ssh_hook,
  64. queue='hpc_4G',
  65. command = 'bash /home/igf/git_repos/IGF-cron-scripts/woolf/woolf_disk_usage.sh '
  66. )
  67. check_igf_lims_disk = \
  68. SSHOperator(
  69. task_id = 'check_igf_lims_disk',
  70. dag = dag,
  71. ssh_hook = igf_lims_ssh_hook,
  72. queue='hpc_4G',
  73. command = 'bash /home/igf/github/IGF-cron-scripts/igf_lims/igf_lims_disk_usage.sh '
  74. )
  75. merge_disk_usage = \
  76. SSHOperator(
  77. task_id = 'merge_disk_usage',
  78. dag = dag,
  79. ssh_hook = eliot_ssh_hook,
  80. queue='hpc_4G',
  81. command = 'bash /home/igf/git_repos/IGF-cron-scripts/eliot/merge_disk_usage.sh '
  82. )
  83. internal_usage = \
  84. SSHOperator(
  85. task_id = 'internal_usage',
  86. dag = dag,
  87. ssh_hook = eliot_ssh_hook,
  88. queue='hpc_4G',
  89. command = 'bash /home/igf/git_repos/IGF-cron-scripts/eliot/internal_usage.sh '
  90. )
  91. merge_disk_usage << [check_orwell_disk,check_eliot_disk,check_woolf_disk,check_igf_lims_disk]
  92. merge_disk_usage >> internal_usage