[Gluster-devel] 2.0.0rc7 and memory leakage

Brent A Nelson brent at phys.ufl.edu
Fri Mar 27 18:26:38 UTC 2009


Is there any known memory leak in the 2.0.0rc7 client? My server processes 
are nicely regulated, but my client memory usage is as high as 6.5GB.  My 
clients are all io-threads=>distribute=>replicate=>client, no other 
translators.

The clients were NFS reexported via unfs3 (I did some 
nfs-kernel-server, but I was getting ACL-related "Operation not supported" 
issues, even with noacl mounts).  Load was produced by rsyncing data (a 
mix of large files and many small files) from remote machines to the NFS.

Note that one of my servers did experience a machine check exception, 
knocking it offline;  I do not know if the excess memory consumption 
occurred before or after that (or, more likely, both).

Thanks,

Brent





More information about the Gluster-devel mailing list