My Oracle Support Banner

ZFS Write Performance Degrades Because All Writes Go to One VDEV (Doc ID 1405345.1)

Last updated on JANUARY 29, 2022

Applies to:

Oracle Communications Messaging Server - Version 6.0.0 and later
Oracle Solaris on SPARC (64-bit)
Oracle Solaris on x86-64 (64-bit)

Symptoms

The symptoms can vary, depending on which components are using the ZFS pool having the problem:

  • The MTA may be slow to accept new incoming SMTP or LMTP connections, because MAX_PROCS processes are already busy with existing connections, which are taking longer than normal.
  • There may be a backlog of messages in the ims-ms or LMTP channels because it is taking longer than normal to insert them into the store.
  • IMAP/POP/Convergence response time may be effected.
  • txn_checkpoint or imdbverify messages in the default log file as in Doc ID 1328634.1.

Changes

A new virtual device (VDEV) was added to the ZFS pool to provide more space.

Cause

To view full details, sign in with your My Oracle Support account.

Don't have a My Oracle Support account? Click to get started!


In this Document
Symptoms
Changes
Cause
 The long-term average stats:
 Current stats:
Solution
 To avoid the problem in the first place
 Will it eventually heal itself?
 Two types of files in the message store
 Tuning to allow the old vdevs to be used
 Results After tuning
 Long-term stats
 Current stats
References


My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.