Ceph » CephFS

Passes:
snaptest-1
trivial_sync
dbench
snaptest-2
bonnie++ (maybe, basic logging doesn't have much output there, will check individually)
snaptest-0
kernel-untar-build (assuming "WARNING: modpost: Found 1 section mismatch(es).
To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'" is in fact a config error and not some weird data bug)
pjd

Fails:
fsx
untar-snap-rm. Gets "cp: cannot stat `.snap/k/coreutils-8.5/.x-sc_useless_cpp_parens': Input/output error" and similar eventually

unknown:
iozone, not installed on pudgy just yet

Now passes fsx.
Closer examination has revealed it fails bonnie++ due to the wrong number of available files, which Sage is looking at.
I think untar-snap-rm might work properly now too, checking it out.

This is now dependent on #383.
And it does pass untar-snap-rm, so it looks like just the readdir issue needs to be fixed.

I tried running the tests on flab and cfuse crashed in one of the snapshot tests. Haven't looked at it much but it tried to dereference a null dentry; it might be something to do with a root-directory snapshot. Or it might not. I'll look into it more today.

Okay, dealt with the crash.
Unfortunately there are other issues in snaptest-2 now, as deleting turns up some ROFS errors and a lot of Dir not empty errors. Assume this came from recent dirfrag changes, will investigate.

It occurs to me that we should also be running these tests on the testing branch, not just unstable. Is that what you've been doing?

Nope, I've been running on unstable. Unfortunately I just hit a crash on bonnie which looks to be new stuff, so now I'm switching to the testing branch so we can at least make sure it works for .21.3.

At this point the only test it's failing is bonnie. This one tends to fail on a SEGV that just keeps going through the backtrace handler until it runs out of memory and gets killed, which makes diagnosing the problem difficult. :/