filterfs: don't use /dev in test#1848
Conversation
hdonnay
force-pushed
the
bug/filterfs-concurrent
branch
from
f2205c0 to
9c530b4
Compare
|
Ah, yeah, I'm not sure this is testing the right thing now. Will rework a bit. |
hdonnay
force-pushed
the
bug/filterfs-concurrent
branch
from
9c530b4 to
39499d2
Compare
|
Updated:
|
|
@crozzy another look, please |
BradLugo
left a comment
BradLugo
left a comment
There was a problem hiding this comment.
LGTM, but I'd like to hear what Crozzy thinks. One minor thing: is it possible to log when an entry is filtered out? If so, what would you think about adding that?
|
It'd be odd because we'd have to smuggle a |
|
Yeah... that's crossed my mind as well. I think it might be nice to plumb |
crozzy
left a comment
crozzy
left a comment
There was a problem hiding this comment.
It lgtm, slices.DeleteFunc is a nice fit. Logging removed files could be nice for debugging but can be discussed/implemented outside of this PR
The `TestDev` test used the machine's `/dev` directory to test special files getting filtered out. However when running concurrently with tests starting and stopping PostgreSQL, the test would flap on `shm` files disappearing from underneath it. This replaces the test with creating a socket in a test-created directory, which will not have files change unexpectedly. Signed-off-by: Hank Donnay <hdonnay@redhat.com> Change-Id: I43f94ee1f59234a423dfead4bccddbfb6a6a6964
hdonnay
force-pushed
the
bug/filterfs-concurrent
branch
from
39499d2 to
f99e8f9
Compare
|
/fast-forward |
3 participants
The
TestDevtest used the machine's/devdirectory to test special files getting filtered out. However when running concurrently with tests starting and stopping PostgreSQL, the test would flap onshmfiles disappearing from underneath it.This replaces the test with creating a fifo in a test-created directory, which will not have files change unexpectedly.