4.1 create_session: Skip test of CB_NULL for nfs4.1
Subsystems (2)
PATCH 4/4 4.1 create_session: Skip test of CB_NULL for nfs4.1
--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
nfs4.1/server41tests/st_create_session.py +2 -2
@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event()
@@ -360,7 +360,7 @@ def testCallbackProgram(t, env): def testCallbackVersion(t, env): """Check server sends callback program with a version listed in nfs4client.py - FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event()
Signature · 1 line
2.4.3
Why are we removing these two test cases? The Ganesha NFS server at least passes them.
Frank
Quoted 44 lines
-----Original Message-----
From: [email protected] [mailto:linux-nfs- [email protected]] On Behalf Of Kinglong Mee
Sent: Tuesday, July 28, 2015 4:47 AM
To: J. Bruce Fields; [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH 4/4] 4.1create_session: Skip test ofCB_NULLfor nfs4.1Signed-off-by Kinglong Mee <[email protected]>--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)nfs4.1/server41tests/st_create_session.py +1
-1@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event()@@ -360,7 +360,7 @@ def testCallbackProgram(t, env): def testCallbackVersion(t, env): """Check server sends callback program with a version listed in
nfs4client.pynfs4.1/server41tests/st_create_session.py +1
-2- FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event() --2.4.3
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected] More majordomo info at
http://vger.kernel.org/majordomo-info.html
Quoted 1 line · Frank Filz
Why are we removing these two test cases? The Ganesha NFS server at least passes them.
And there's nothing wrong with passing the tests, but unfortunately there's not necessarily anything wrong with failing either: a 4.1 server isn't required to do a NULL callback to probe the backchannel.
There might be a better test (depend on delegation recalls instead of CB_NULL's?), till then I'm OK with turning them off by default.
--b.
Quoted 47 lines
Frank
Quoted 44 lines
-----Original Message-----
From: [email protected] [mailto:linux-nfs- [email protected]] On Behalf Of Kinglong Mee
Sent: Tuesday, July 28, 2015 4:47 AM
To: J. Bruce Fields; [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH 4/4] 4.1create_session: Skip test ofCB_NULLfor nfs4.1Signed-off-by Kinglong Mee <[email protected]>--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)nfs4.1/server41tests/st_create_session.py +1
-1@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event()@@ -360,7 +360,7 @@ def testCallbackProgram(t, env): def testCallbackVersion(t, env): """Check server sends callback program with a version listed in
nfs4client.pynfs4.1/server41tests/st_create_session.py +1
-2- FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event() --2.4.3
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected] More majordomo info at
http://vger.kernel.org/majordomo-info.html
Quoted 5 lines · Frank Filz
Quoted 1 line
Why are we removing these two test cases? The Ganesha NFS server at
least passes them.
And there's nothing wrong with passing the tests, but unfortunately
there's
Quoted 1 line
not necessarily anything wrong with failing either: a 4.1 server isn't
required
Quoted 4 lines
to do a NULL callback to probe the backchannel.
There might be a better test (depend on delegation recalls instead of CB_NULL's?), till then I'm OK with turning them off by default.
I guess I'd at least like to see some explanation of why we're turning the tests off by default. I'm just resistant to removing testing for no reason.
If the test doesn't actually test anything useful, that would be one explanation.
If it's something that's not required and not all servers do it, then that's another explanation.
There might be other explanations, just please give some insight.
Thanks
Frank
Quoted 41 lines
Quoted 41 lines
Quoted 41 lines
-----Original Message-----
From: [email protected] [mailto:linux-nfs- [email protected]] On Behalf Of Kinglong Mee
Sent: Tuesday, July 28, 2015 4:47 AM
To: J. Bruce Fields; [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH 4/4] 4.1create_session: Skip test ofCB_NULLfor nfs4.1Signed-off-by Kinglong Mee <[email protected]>--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)nfs4.1/server41tests/st_create_session.py +1
-1@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event() @@ -360,7 +360,7 @@ def
testCallbackProgram(t, env):
deftestCallbackVersion(t, env):
"""Check server sends callback program with a version listed innfs4client.pynfs4.1/server41tests/st_create_session.py +1
-2- FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event() --2.4.3
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected] Moremajordomo
Quoted 1 line
Quoted 1 line
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the
Quoted 2 lines
body of a message to [email protected] More majordomo info at
http://vger.kernel.org/majordomo-info.html
Quoted 7 lines · Frank Filz
Quoted 5 lines · Frank Filz
Quoted 1 line
Why are we removing these two test cases? The Ganesha NFS server at
least passes them.
And there's nothing wrong with passing the tests, but unfortunately
there's
Quoted 1 line
not necessarily anything wrong with failing either: a 4.1 server isn't
Quoted 8 lines
required
Quoted 4 lines
to do a NULL callback to probe the backchannel.
There might be a better test (depend on delegation recalls instead of CB_NULL's?), till then I'm OK with turning them off by default.
I guess I'd at least like to see some explanation of why we're turning the tests off by default. I'm just resistant to removing testing for no reason.
The story is: we turned off the CB_NULL probe in knfsd in the 4.1 case; it's unnecessary, and it was causing some problems (due partly to other bugs, but why ask for trouble?). So these tests started failing. I'd fix the server if there was an actual bug, but it's the tests that are wrong in this case, so oh well.
Quoted 6 lines
If the test doesn't actually test anything useful, that would be one explanation.
If it's something that's not required and not all servers do it, then that's another explanation.
Right, this is it, CB_NULL probes aren't required.
We're testing something useful, but something you need a callback to test, so trying to get some other kind of callback (like a deleg recall) would be another option.
--b.
Quoted 58 lines
There might be other explanations, just please give some insight.
Thanks
Frank
Quoted 41 lines
Quoted 41 lines
Quoted 41 lines
-----Original Message-----
From: [email protected] [mailto:linux-nfs- [email protected]] On Behalf Of Kinglong Mee
Sent: Tuesday, July 28, 2015 4:47 AM
To: J. Bruce Fields; [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH 4/4] 4.1create_session: Skip test ofCB_NULLfor nfs4.1Signed-off-by Kinglong Mee <[email protected]>--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)nfs4.1/server41tests/st_create_session.py +1
-1@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event() @@ -360,7 +360,7 @@ def
testCallbackProgram(t, env):
deftestCallbackVersion(t, env):
"""Check server sends callback program with a version listed innfs4client.pynfs4.1/server41tests/st_create_session.py +1
-2- FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event() --2.4.3
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to [email protected] Moremajordomo
Quoted 1 line
Quoted 1 line
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" inthe
Quoted 2 lines
body of a message to [email protected] More majordomo info at
http://vger.kernel.org/majordomo-info.html
Quoted 10 lines · Frank Filz
Quoted 9 lines
Quoted 6 lines · Frank Filz
Quoted 2 lines
Why are we removing these two test cases? The Ganesha NFS server at
least passes them.
And there's nothing wrong with passing the tests, but unfortunately
there's
Quoted 2 lines
not necessarily anything wrong with failing either: a 4.1 server isn't
Quoted 8 lines
Quoted 8 lines
required
Quoted 4 lines
to do a NULL callback to probe the backchannel.
There might be a better test (depend on delegation recalls instead of CB_NULL's?), till then I'm OK with turning them off by default.
I guess I'd at least like to see some explanation of why we're turning the tests off by default. I'm just resistant to removing testing for no
reason.
Quoted 2 lines
The story is: we turned off the
CB_NULLprobe in knfsd in the 4.1 case;
it's
Quoted 2 lines
unnecessary, and it was causing some problems (due partly to other bugs, but why ask for trouble?). So these tests started failing. I'd fix the
server if
Quoted 1 line
there was an actual bug, but it's the tests that are wrong in this case,
so oh
Quoted 1 line
well.
Ah, ok. I'd love to understand the issues there better, maybe Ganesha shouldn't be doing the CB_NULL either...
Quoted 7 lines
Quoted 5 lines
If the test doesn't actually test anything useful, that would be one explanation.
If it's something that's not required and not all servers do it, then that's another explanation.
Right, this is it,
CB_NULLprobes aren't required.
Ok, with this perspective, I'm cool with the test being removed from default.
Quoted 2 lines
We're testing something useful, but something you need a callback to test, so trying to get some other kind of callback (like a deleg recall) would
be
Quoted 1 line
another option.
Hmm, more to think about from testing Ganesha perspective. The "simple" thing (i.e. using FSAL_VFS) to test in Ganesha doesn't do delegations, so the CB_NULL might be the only callback it makes, which might make it somewhat useful to prove Ganesha's callback code isn't totally broken...
Thanks for the additional explanation.
Frank
Quoted 48 lines
Quoted 48 lines
There might be other explanations, just please give some insight.
Thanks
Frank
Quoted 42 lines
Quoted 42 lines
Quoted 42 lines
-----Original Message-----
From: [email protected] [mailto:linux-nfs- [email protected]] On Behalf Of Kinglong Mee
Sent: Tuesday, July 28, 2015 4:47 AM
To: J. Bruce Fields; [email protected]
Cc: [email protected]; [email protected]
Subject: [PATCH 4/4] 4.1create_session: Skip test ofCB_NULLfor
nfs4.1Signed-off-by Kinglong Mee <[email protected]>--- nfs4.1/server41tests/st_create_session.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)nfs4.1/server41tests/st_create_session.py +1
-1@@ -333,7 +333,7 @@ def testManyClients(t, env): def testCallbackProgram(t, env): """Check server can handle random transient program number - FLAGS: create_session all + FLAGS: CODE: CSESS20 """ cb_occurred = threading.Event() @@ -360,7 +360,7 @@ def
testCallbackProgram(t, env):
deftestCallbackVersion(t, env):
"""Check server sends callback program with a version
listed innfs4client.pynfs4.1/server41tests/st_create_session.py +1
-2- FLAGS: create_session all + FLAGS: CODE: CSESS21 """ cb_occurred = threading.Event() --2.4.3
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
--
To unsubscribe from this list: send the line "unsubscribe
linux-nfs"
Quoted 10 lines
Quoted 8 lines
Quoted 6 lines
Quoted 1 line
Quoted 1 line
in the body of a message to [email protected] More
majordomo
Quoted 1 line
Quoted 1 line
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" inthe
Quoted 1 line
body of a message to [email protected] More majordomo
info
Quoted 1 line
Quoted 1 line
Quoted 24 lines · Frank Filz
Quoted 10 lines · Frank Filz
Quoted 9 lines
Quoted 6 lines · Frank Filz
Quoted 2 lines
Why are we removing these two test cases? The Ganesha NFS server at
least passes them.
And there's nothing wrong with passing the tests, but unfortunately
there's
Quoted 2 lines
not necessarily anything wrong with failing either: a 4.1 server isn't
Quoted 8 lines
Quoted 8 lines
required
Quoted 4 lines
to do a NULL callback to probe the backchannel.
There might be a better test (depend on delegation recalls instead of CB_NULL's?), till then I'm OK with turning them off by default.
I guess I'd at least like to see some explanation of why we're turning the tests off by default. I'm just resistant to removing testing for no
reason.
Quoted 2 lines
The story is: we turned off the
CB_NULLprobe in knfsd in the 4.1 case;it's
Quoted 2 lines
unnecessary, and it was causing some problems (due partly to other bugs, but why ask for trouble?). So these tests started failing. I'd fix the
Quoted 7 lines
server if
Quoted 1 line
there was an actual bug, but it's the tests that are wrong in this case,
so oh
Quoted 1 line
well.
Ah, ok. I'd love to understand the issues there better, maybe Ganesha shouldn't be doing the
CB_NULLeither...
I've actually forgotten; the patches are from April so linux-nfs archives from around there might have some discussion. I think there was actually a bug in client callback handling that frequent unnecessary CB_NULL's were triggering.
Quoted 13 lines
Quoted 7 lines
Quoted 5 lines
If the test doesn't actually test anything useful, that would be one explanation.
If it's something that's not required and not all servers do it, then that's another explanation.
Right, this is it,
CB_NULLprobes aren't required.Ok, with this perspective, I'm cool with the test being removed from default.
Quoted 2 lines
We're testing something useful, but something you need a callback to test, so trying to get some other kind of callback (like a deleg recall) would
Quoted 7 lines
be
Quoted 1 line
another option.
Hmm, more to think about from testing Ganesha perspective. The "simple" thing (i.e. using
FSAL_VFS) to test in Ganesha doesn't do delegations, so theCB_NULLmight be the only callback it makes, which might make it somewhat useful to prove Ganesha's callback code isn't totally broken...
Yeah, for testing it's annoying that there's no reliable to provoke a callback. pynfs could probably be smarter about trying various methods, and also about reporting the difference between "X failed" and "I couldn't test X because I couldn't get server to optional thing Y".
--b.