Oracle HSM - archiver "Fatal OS call error" - ZFS Memory Usage Needing Adjustment
(Doc ID 2044949.1)
Last updated on MAY 31, 2018
Applies to:Oracle Hierarchical Storage Manager (HSM) and StorageTek QFS Software - Version 5.4 and later
Information in this document applies to any platform.
The customer is running a heavily loaded SAM-QFS 5.4 metadata server on a T5-2 host with 256 GB RAM and Solaris 11.1. The memory usage starts out at about 32 GB, but gradually and in big jumps, reaches ~250 GB, at which point SAM-FS starts issuing /var/opt/SUNWsamfs/trace/archiverd messages as follows:
sam-arcopy[4370:1]: cirbuf.c:165 Fatal OS call error: mlock(cb_buf) called from: cirbuf.c:165: Resource temporarily unavailable
sam-archiverd[26673:2]: archiverd.c:1378 Message 4027: sam-arcoipy exit status 0300
Looking at ps output and pmap -x for each process does not show any significant amount of memory in use. These messages began showing up after upgrading to SAMFS 5.4 and Solaris 11.1 (at about the same time). All the tape types have 2 MB block sizes and 64 "locked" blocks specified in archiver.cmd and stager.cmd. We have 56 tape drives, so about 56*128MB ~ 7.2 GB could be accounted for by staging and archiving buffers, which is far short of the 250 GB in use. The block sizes and the other background activity were not changed from the older configuration, which ran SAM-QFS 5.3.14 on Solaris 10.
Upgraded from SAM-QFS 5.3.14 on Solaris 10, to SAMFS 5.4 and Solaris 11.1
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