IJ49701: PROCESSES HANG DUE TO DEADLOCKS IN OUR STORAGE SCALE CLUSTER

APAR status

• Closed as program error.

Error description

• Processes hang due to deadlocks in our Storage Scale cluster.
  There aredeadlock notifications on multiple nodes which were
  triggered by 'long waiter' events on the nodes
  

Local fix

Problem summary

• Processes hang due to deadlocks in our Storage Scale cluster.
  There aredeadlock notifications on multiple nodes which were
  triggered by 'long waiter' events on the nodes
  

Problem conclusion

• This problem is fixed in 5.1.9.2
  To see all Spectrum Scale APARs and their respective
  Fix solutions refer to page:
  https://public.dhe.ibm.com/storage/spectrumscale/spectrum_scale_
  apars.html
  
  Benefits of the solution:
  Newer Linux kernels try to determine if performing a readahead
  would be beneficial and trigger it if so. This creates certain
  challenges for distributed filesystems which often ends up in
  locking related issues. The solution implements the readahead
  VFS operation to handle this issue.
  
  Work Around:
  None
  
  Problem trigger:
  A single large file being read sequentially from one
  node(causing a readahead to be performed on the file or by using
  a posix_fadvise call to trigger readahead forcefully) and also
  being truncated/deleted from another node at the same
  time.
  
  Symptom:
  Client processes hang and system deadlocks
  
  Platforms affected:
  Linux Only
  
  Functional Area affected:
  Regular file read flow in kernel version >= 5.14
  
  Customer Impact:
  High Importance
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  SPEC SCALE STD

• Fixed component ID

  5737F33AP

Applicable component levels