| Message ID | 20190716122836.31187-3-thuth@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Enable iotests during "make check" | expand |
On 16.07.19 14:28, Thomas Huth wrote: > Remove some more tests from the "auto" group that either have issues > in certain environments (like macOS or FreeBSD, or on certain file systems > like ZFS or tmpfs), do not work with the qcow2 format, or that are simply > taking too much time. > > Signed-off-by: Thomas Huth <thuth@redhat.com> > --- > tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- > 1 file changed, 58 insertions(+), 56 deletions(-) I just looked through the list to see whether any of the test seems like we’d want to keep it even though it is a bit slow. Mostly I was looking for tests that cover complex cases. 255 is the only one that seemed to fit that bill to me. So why do you remove it? Is it because it takes two seconds? Max
On 16/07/2019 17.26, Max Reitz wrote: > On 16.07.19 14:28, Thomas Huth wrote: >> Remove some more tests from the "auto" group that either have issues >> in certain environments (like macOS or FreeBSD, or on certain file systems >> like ZFS or tmpfs), do not work with the qcow2 format, or that are simply >> taking too much time. >> >> Signed-off-by: Thomas Huth <thuth@redhat.com> >> --- >> tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- >> 1 file changed, 58 insertions(+), 56 deletions(-) > > I just looked through the list to see whether any of the test seems like > we’d want to keep it even though it is a bit slow. Mostly I was looking > for tests that cover complex cases. > > 255 is the only one that seemed to fit that bill to me. So why do you > remove it? Is it because it takes two seconds? No, I removed it because it was failing on macOS: https://cirrus-ci.com/task/4860239294234624 ("OSError: AF_UNIX path too long" is the error, if I got that right) Thomas
On 16.07.19 17:31, Thomas Huth wrote: > On 16/07/2019 17.26, Max Reitz wrote: >> On 16.07.19 14:28, Thomas Huth wrote: >>> Remove some more tests from the "auto" group that either have issues >>> in certain environments (like macOS or FreeBSD, or on certain file systems >>> like ZFS or tmpfs), do not work with the qcow2 format, or that are simply >>> taking too much time. >>> >>> Signed-off-by: Thomas Huth <thuth@redhat.com> >>> --- >>> tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- >>> 1 file changed, 58 insertions(+), 56 deletions(-) >> >> I just looked through the list to see whether any of the test seems like >> we’d want to keep it even though it is a bit slow. Mostly I was looking >> for tests that cover complex cases. >> >> 255 is the only one that seemed to fit that bill to me. So why do you >> remove it? Is it because it takes two seconds? > > No, I removed it because it was failing on macOS: > > https://cirrus-ci.com/task/4860239294234624 > > ("OSError: AF_UNIX path too long" is the error, if I got that right) Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? Max
On 16.07.19 17:44, Max Reitz wrote: > On 16.07.19 17:31, Thomas Huth wrote: >> On 16/07/2019 17.26, Max Reitz wrote: >>> On 16.07.19 14:28, Thomas Huth wrote: >>>> Remove some more tests from the "auto" group that either have issues >>>> in certain environments (like macOS or FreeBSD, or on certain file systems >>>> like ZFS or tmpfs), do not work with the qcow2 format, or that are simply >>>> taking too much time. >>>> >>>> Signed-off-by: Thomas Huth <thuth@redhat.com> >>>> --- >>>> tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- >>>> 1 file changed, 58 insertions(+), 56 deletions(-) >>> >>> I just looked through the list to see whether any of the test seems like >>> we’d want to keep it even though it is a bit slow. Mostly I was looking >>> for tests that cover complex cases. >>> >>> 255 is the only one that seemed to fit that bill to me. So why do you >>> remove it? Is it because it takes two seconds? >> >> No, I removed it because it was failing on macOS: >> >> https://cirrus-ci.com/task/4860239294234624 >> >> ("OSError: AF_UNIX path too long" is the error, if I got that right) > > Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? (Forgot the Reviewed-by: Max Reitz <mreitz@redhat.com> because I’m so... amazed by that possibility)
On 7/16/19 10:44 AM, Max Reitz wrote: > On 16.07.19 17:31, Thomas Huth wrote: >> On 16/07/2019 17.26, Max Reitz wrote: >>> On 16.07.19 14:28, Thomas Huth wrote: >>>> Remove some more tests from the "auto" group that either have issues >>>> in certain environments (like macOS or FreeBSD, or on certain file systems >>>> like ZFS or tmpfs), do not work with the qcow2 format, or that are simply >>>> taking too much time. >>>> >>>> Signed-off-by: Thomas Huth <thuth@redhat.com> >>>> --- >>>> tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- >>>> 1 file changed, 58 insertions(+), 56 deletions(-) >>> >>> I just looked through the list to see whether any of the test seems like >>> we’d want to keep it even though it is a bit slow. Mostly I was looking >>> for tests that cover complex cases. >>> >>> 255 is the only one that seemed to fit that bill to me. So why do you >>> remove it? Is it because it takes two seconds? >> >> No, I removed it because it was failing on macOS: >> >> https://cirrus-ci.com/task/4860239294234624 >> >> ("OSError: AF_UNIX path too long" is the error, if I got that right) > > Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? Not when our CI is set up to use super-long file names: + File "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/../../python/qemu/machine.py", line 294, in launch Is there any way to create our sockets somewhere under /tmp instead of inside tests/qemu-iotests, so that we have a shorter filename for sockets no matter how deep in the file hierarchy the tests themselves live? Also, at one point, we tossed around the idea of s/qemu-iotests/iotests/, to shave off 5 characters that don't really add anything.
On 16.07.19 17:56, Eric Blake wrote: > On 7/16/19 10:44 AM, Max Reitz wrote: >> On 16.07.19 17:31, Thomas Huth wrote: >>> On 16/07/2019 17.26, Max Reitz wrote: >>>> On 16.07.19 14:28, Thomas Huth wrote: >>>>> Remove some more tests from the "auto" group that either have issues >>>>> in certain environments (like macOS or FreeBSD, or on certain file systems >>>>> like ZFS or tmpfs), do not work with the qcow2 format, or that are simply >>>>> taking too much time. >>>>> >>>>> Signed-off-by: Thomas Huth <thuth@redhat.com> >>>>> --- >>>>> tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- >>>>> 1 file changed, 58 insertions(+), 56 deletions(-) >>>> >>>> I just looked through the list to see whether any of the test seems like >>>> we’d want to keep it even though it is a bit slow. Mostly I was looking >>>> for tests that cover complex cases. >>>> >>>> 255 is the only one that seemed to fit that bill to me. So why do you >>>> remove it? Is it because it takes two seconds? >>> >>> No, I removed it because it was failing on macOS: >>> >>> https://cirrus-ci.com/task/4860239294234624 >>> >>> ("OSError: AF_UNIX path too long" is the error, if I got that right) >> >> Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? > > Not when our CI is set up to use super-long file names: > > + File > "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/../../python/qemu/machine.py", > line 294, in launch That isn’t really long. > Is there any way to create our sockets somewhere under /tmp instead of > inside tests/qemu-iotests, so that we have a shorter filename for > sockets no matter how deep in the file hierarchy the tests themselves live? > > Also, at one point, we tossed around the idea of > s/qemu-iotests/iotests/, to shave off 5 characters that don't really add > anything. I’d personally rather just skip the iotests if we detect such a silly OS, but maybe that’s just me. Max
On 7/16/19 10:58 AM, Max Reitz wrote: >>> Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? >> >> Not when our CI is set up to use super-long file names: >> >> + File >> "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/../../python/qemu/machine.py", >> line 294, in launch > > That isn’t really long. The MacOS limit is 104 (or 103 if you insist on using the trailing NUL yourself), compared to Linux at 108 (107). And: $ printf "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/scratch/name" | wc -c 104 shows that we are right on the verge of overflowing the space allotted, with any socket name worth using when the socket lives inside qemu-iotests/scratch. > >> Is there any way to create our sockets somewhere under /tmp instead of >> inside tests/qemu-iotests, so that we have a shorter filename for >> sockets no matter how deep in the file hierarchy the tests themselves live? >> >> Also, at one point, we tossed around the idea of >> s/qemu-iotests/iotests/, to shave off 5 characters that don't really add >> anything. > > I’d personally rather just skip the iotests if we detect such a silly > OS, but maybe that’s just me. It's a rather unfortunate limit, but it's not all that silly (the limit is based on the fact that struct sockaddr has to fit inside a nice power-of-2 structure somewhere in the kernel, and enough else is used that you really are left with just 104/108 bytes, or even 92 bytes if you use HP-UX 11). POSIX does not place a minimum length on sun_path, but I know of no system that does not allow at least 92 bytes, if you are aiming for a portably-small name.
On 16.07.19 18:09, Eric Blake wrote: > On 7/16/19 10:58 AM, Max Reitz wrote: > >>>> Ah, OK. So, uh, we effectively can’t run any Python tests on macOS? >>> >>> Not when our CI is set up to use super-long file names: >>> >>> + File >>> "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/../../python/qemu/machine.py", >>> line 294, in launch >> >> That isn’t really long. > > The MacOS limit is 104 (or 103 if you insist on using the trailing NUL > yourself), compared to Linux at 108 (107). And: > > $ printf > "/private/var/folders/sy/2x5qvs0n4lg18fry9jz4y21m0000gn/T/cirrus-ci-build/tests/qemu-iotests/scratch/name" > | wc -c > 104 > > shows that we are right on the verge of overflowing the space allotted, > with any socket name worth using when the socket lives inside > qemu-iotests/scratch. > >> >>> Is there any way to create our sockets somewhere under /tmp instead of >>> inside tests/qemu-iotests, so that we have a shorter filename for >>> sockets no matter how deep in the file hierarchy the tests themselves live? >>> >>> Also, at one point, we tossed around the idea of >>> s/qemu-iotests/iotests/, to shave off 5 characters that don't really add >>> anything. >> >> I’d personally rather just skip the iotests if we detect such a silly >> OS, but maybe that’s just me. > > It's a rather unfortunate limit, but it's not all that silly (the limit > is based on the fact that struct sockaddr has to fit inside a nice > power-of-2 structure somewhere in the kernel, and enough else is used > that you really are left with just 104/108 bytes, or even 92 bytes if > you use HP-UX 11). POSIX does not place a minimum length on sun_path, > but I know of no system that does not allow at least 92 bytes, if you > are aiming for a portably-small name. It does come to me as a surprise that the sockaddr is a path instead of, say, an inode number. But shaving off the “qemu-” seems like the wrong approach to me still. Max
On 16/07/2019 18.12, Max Reitz wrote: > On 16.07.19 18:09, Eric Blake wrote: >> On 7/16/19 10:58 AM, Max Reitz wrote: [...] >>>> Is there any way to create our sockets somewhere under /tmp instead of >>>> inside tests/qemu-iotests, so that we have a shorter filename for >>>> sockets no matter how deep in the file hierarchy the tests themselves live? >>>> >>>> Also, at one point, we tossed around the idea of >>>> s/qemu-iotests/iotests/, to shave off 5 characters that don't really add >>>> anything. >>> >>> I’d personally rather just skip the iotests if we detect such a silly >>> OS, but maybe that’s just me. >> >> It's a rather unfortunate limit, but it's not all that silly (the limit >> is based on the fact that struct sockaddr has to fit inside a nice >> power-of-2 structure somewhere in the kernel, and enough else is used >> that you really are left with just 104/108 bytes, or even 92 bytes if >> you use HP-UX 11). POSIX does not place a minimum length on sun_path, >> but I know of no system that does not allow at least 92 bytes, if you >> are aiming for a portably-small name. > > It does come to me as a surprise that the sockaddr is a path instead of, > say, an inode number. > > But shaving off the “qemu-” seems like the wrong approach to me still. Anyway, it would be great if someone who's familiar with the code could find a proper fix for this problem. FWIW, it also happens on Linux: mkdir /tmp/0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789 cd /tmp/0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789/ ~/path/to/qemu/configure --target-list=x86_64-softmmu make -j8 cd tests/qemu-iotests/ ./check -g quick -qcow2 ... 058 fail [09:29:33] [09:29:34] output mismatch (see 058.out.bad) --- /home/thuth/devel/qemu/tests/qemu-iotests/058.out 2019-04-23 16:43:12.000000000 +0200 +++ /tmp/0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789/tests/qemu-iotests/058.out.bad 2019-07-18 09:29:34.305070819 +0200 @@ -17,18 +17,22 @@ 4 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) read 4096/4096 bytes at offset 8192 4 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +qemu-nbd: UNIX socket path '/tmp/0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789/tests/qemu-iotests/scratch/qemu-nbd.sock' is too long +Path must be less than 108 bytes ... 065 fail [09:29:38] [09:29:38] output mismatch (see 065.out.bad) --- /home/thuth/devel/qemu/tests/qemu-iotests/065.out 2019-04-23 16:43:12.000000000 +0200 +++ /tmp/0123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789/tests/qemu-iotests/065.out.bad 2019-07-18 09:29:38.568118520 +0200 @@ -1,5 +1,41 @@ -........ +....E..E +====================================================================== +ERROR: test_qmp (__main__.TestQCow3LazyQMP) +---------------------------------------------------------------------- +Traceback (most recent call last): + File "065", line 75, in setUp + self.vm.launch() + File "/home/thuth/devel/qemu/tests/qemu-iotests/../../python/qemu/machine.py", line 294, in launch + self._launch() + File "/home/thuth/devel/qemu/tests/qemu-iotests/../../python/qemu/machine.py", line 311, in _launch + self._pre_launch() + File "/home/thuth/devel/qemu/tests/qemu-iotests/../../python/qemu/qtest.py", line 103, in _pre_launch + super(QEMUQtestMachine, self)._pre_launch() + File "/home/thuth/devel/qemu/tests/qemu-iotests/../../python/qemu/machine.py", line 262, in _pre_launch + server=True) + File "/home/thuth/devel/qemu/tests/qemu-iotests/../../python/qemu/qmp.py", line 60, in __init__ + self.__sock.bind(self.__address) +OSError: AF_UNIX path too long ... Thomas
diff --git a/tests/qemu-iotests/group b/tests/qemu-iotests/group index b34c8e3c0c..0ae555b610 100644 --- a/tests/qemu-iotests/group +++ b/tests/qemu-iotests/group @@ -14,7 +14,9 @@ # runnable in any case. That means they should run with every QEMU binary # (also non-x86), with every QEMU configuration (i.e. must not fail if # an optional feature is not compiled in - but reporting a "skip" is ok), -# and work all kind of host filesystems and users (e.g. "nobody" or "root"). +# work at least with the qcow2 file format, work with all kind of host +# filesystems and users (e.g. "nobody" or "root") and must not take too +# much memory and disk space (since CI pipelines tend to fail otherwise). # # @@ -33,8 +35,8 @@ 011 rw auto quick 012 auto quick 013 rw auto -014 rw auto -015 rw snapshot auto +014 rw +015 rw snapshot # 016 was removed, do not reuse 017 rw backing auto quick 018 rw backing auto quick @@ -42,7 +44,7 @@ 020 rw backing auto quick 021 io auto quick 022 rw snapshot auto -023 rw auto +023 rw 024 rw backing auto quick 025 rw auto quick 026 rw blkdbg @@ -78,94 +80,94 @@ 056 rw backing 057 rw 058 rw quick -059 rw auto quick +059 rw quick 060 rw auto quick 061 rw auto 062 rw auto quick 063 rw auto quick -064 rw auto quick +064 rw quick 065 rw quick 066 rw auto quick 067 rw quick 068 rw quick 069 rw auto quick -070 rw auto quick +070 rw quick 071 rw auto quick 072 rw auto quick 073 rw auto quick 074 rw auto quick -075 rw auto quick -076 auto -077 rw auto quick -078 rw auto quick +075 rw quick +076 io +077 rw quick +078 rw quick 079 rw auto 080 rw auto -081 rw auto quick -082 rw auto quick -083 rw auto -084 img auto quick +081 rw quick +082 rw quick +083 rw +084 img quick 085 rw 086 rw auto quick 087 rw quick -088 rw auto quick +088 rw quick 089 rw auto quick 090 rw auto quick -091 rw auto migration -092 rw auto quick +091 rw migration +092 rw quick 093 throttle -094 rw auto quick +094 rw quick 095 rw quick 096 rw quick 097 rw auto backing 098 rw auto backing quick 099 rw auto quick # 100 was removed, do not reuse -101 rw auto quick -102 rw auto quick +101 rw quick +102 rw quick 103 rw auto quick 104 rw auto 105 rw auto quick -106 rw auto quick +106 rw quick 107 rw auto quick 108 rw auto quick -109 rw auto +109 rw 110 rw auto backing quick 111 rw auto quick 112 rw -113 rw auto quick +113 rw quick 114 rw auto quick 115 rw -116 rw auto quick +116 rw quick 117 rw auto 118 rw -119 rw auto quick +119 rw quick 120 rw auto quick 121 rw 122 rw auto -123 rw auto quick +123 rw quick 124 rw backing 125 rw 126 rw auto backing 127 rw backing quick -128 rw auto quick +128 rw quick 129 rw quick 130 rw auto quick -131 rw auto quick +131 rw quick 132 rw quick 133 auto quick 134 rw auto quick -135 rw auto +135 rw 136 rw 137 rw auto 138 rw auto quick 139 rw quick 140 rw auto quick 141 rw auto quick -142 auto +142 143 auto quick 144 rw quick 145 quick -146 auto quick +146 quick 147 img 148 rw quick 149 rw sudo @@ -179,18 +181,18 @@ 157 quick 158 rw auto quick 159 rw auto quick -160 rw auto quick +160 rw quick 161 rw auto quick 162 quick 163 rw 165 rw quick 169 rw quick migration 170 rw auto quick -171 rw auto quick +171 rw quick 172 auto -173 rw auto +173 rw 174 auto -175 auto quick +175 quick 176 rw auto backing 177 rw auto quick 178 img @@ -210,7 +212,7 @@ 194 rw migration quick 195 rw auto quick 196 rw quick migration -197 rw auto quick +197 rw quick 198 rw 199 rw migration 200 rw @@ -220,52 +222,52 @@ 204 rw quick 205 rw quick 206 rw -207 rw auto +207 rw 208 rw quick 209 rw quick -210 rw auto -211 rw auto quick -212 rw auto quick -213 rw auto quick +210 rw +211 rw quick +212 rw quick +213 rw quick 214 rw auto -215 rw auto quick +215 rw quick 216 rw quick 217 rw auto quick 218 rw quick 219 rw 220 rw auto -221 rw auto quick +221 rw quick 222 rw quick 223 rw quick 224 rw quick -225 rw auto quick +225 rw quick 226 auto quick 227 quick 228 rw quick 229 auto quick -231 auto quick +231 quick 232 quick -233 auto quick +233 quick 234 quick migration 235 quick 236 quick -237 rw auto quick +237 rw quick 238 quick -239 rw auto quick +239 rw quick 240 quick -241 rw auto quick +241 rw quick 242 rw quick -243 rw auto quick +243 rw quick 244 rw auto quick 245 rw 246 rw quick 247 rw quick 248 rw quick 249 rw auto quick -250 rw auto quick +250 rw quick 251 rw auto quick 252 rw auto backing quick -253 rw auto quick -254 rw auto backing quick -255 rw auto quick -256 rw auto quick +253 rw quick +254 rw backing quick +255 rw quick +256 rw quick
Remove some more tests from the "auto" group that either have issues in certain environments (like macOS or FreeBSD, or on certain file systems like ZFS or tmpfs), do not work with the qcow2 format, or that are simply taking too much time. Signed-off-by: Thomas Huth <thuth@redhat.com> --- tests/qemu-iotests/group | 114 ++++++++++++++++++++------------------- 1 file changed, 58 insertions(+), 56 deletions(-)