[Gluster-devel] Memory Leak Strikes Again

[Gordan Bobic]

I have just noticed that the memory leak on the root file system appears 
to still be occuring. The only things I have changed are that I have now 
removed the "direct-io-mode=off,log-level=NONE,log-file=/dev/null" mount 
parameters, so it appears to be related. The process grows to over 350MB 
during a single test pass.

The operation that causes major leakage is making the initrd for Open 
Shared Root (much more involved than normal mkinitrd, as it makes 
repeated calls to the rpm database, which, in this instance, is in 
SQLite format because BDB flatly refuses to work on GlusterFS). Leakage 
is approximately 1MB every 5 seconds during the operation, and GlusterFS 
uses 100% of CPU while the operation is occurring (which is in itself a 
tad worrying even if it is running on a 1.6GHz Atom 330, which is not 
exactly an underpowered machine). rootfs is AFR, but the machine is 
operating standalone (no peers connected).

I have just put it back as it was (mounted with
-o direct-io-mode=off,log-level=NONE,log-file=/dev/null
and sure enough the leak disappeared. The CPU usage is also down to a 
much more reasonable 25-30%.

Should I bother with a bugzilla report, or is this already a known issue 
in 2.0.2?

Gordan