1 _______________________
2 BUILDING THE FSQA SUITE
3 _______________________
6 - cd into the xfstests directory
7 - install prerequisite packages
8 For example, for Ubuntu:
9 sudo apt-get install xfslibs-dev uuid-dev libtool-bin \
10 e2fsprogs automake gcc libuuid1 quota attr libattr1-dev make \
11 libacl1-dev libaio-dev xfsprogs libgdbm-dev gawk fio dbench \
13 For Fedora, RHEL, or CentOS:
14 yum install acl attr automake bc dbench dump e2fsprogs fio \
15 gawk gcc indent libtool lvm2 make psmisc quota sed \
17 libacl-devel libattr-devel libaio-devel libuuid-devel \
18 xfsprogs-devel btrfs-progs-devel
19 (Older distributions may require xfsprogs-qa-devel as well.)
20 (Note that for RHEL and CentOS, you may need the EPEL repo.)
23 - create fsgqa test user ("sudo useradd fsgqa")
24 - create 123456-fsgqa test user ("sudo useradd 123456-fsgqa")
27 - cd into the xfstests directory
28 - set the ROOT and TOOLROOT env variables for IRIX appropriately
31 ______________________
33 ______________________
35 Preparing system for tests (IRIX and Linux):
37 - compile XFS into your kernel or load XFS modules
38 - install administrative tools specific to the filesystem you wish to test
39 - If you wish to run the udf components of the suite install
40 mkfs_udf and udf_db for IRIX and mkudffs for Linux. Also download and
41 build the Philips UDF Verification Software from
42 http://www.extra.research.philips.com/udf/, then copy the udf_test
43 binary to xfstests/src/. If you wish to disable UDF verification test
44 set the environment variable DISABLE_UDF_TEST to 1.
47 - create one or two partitions to use for testing
49 - format as XFS, mount & optionally populate with
51 - one SCRATCH partition (optional)
52 - leave empty and expect this partition to be clobbered
53 by some tests. If this is not provided, many tests will
55 (SCRATCH and TEST must be two DIFFERENT partitions)
57 - for btrfs only: some btrfs test cases will need 3 or more independent
58 SCRATCH disks which should be set using SCRATCH_DEV_POOL (for eg:
59 SCRATCH_DEV_POOL="/dev/sda /dev/sdb /dev/sdc") with which
60 SCRATCH_DEV should be unused by the tester, and for the legacy
61 support SCRATCH_DEV will be set to the first disk of the
62 SCRATCH_DEV_POOL by xfstests script.
64 - setup your environment
66 - copy local.config.example to local.config and edit as needed
68 - setenv TEST_DEV "device containing TEST PARTITION"
69 - setenv TEST_DIR "mount point of TEST PARTITION"
71 - setenv SCRATCH_DEV "device containing SCRATCH PARTITION" OR
72 (btrfs only) setenv SCRATCH_DEV_POOL "to 3 or more SCRATCH disks for
73 testing btrfs raid concepts"
74 - setenv SCRATCH_MNT "mount point for SCRATCH PARTITION"
75 - setenv TAPE_DEV "tape device for testing xfsdump"
76 - setenv RMT_TAPE_DEV "remote tape device for testing xfsdump"
77 - setenv RMT_IRIXTAPE_DEV "remote IRIX tape device for testing xfsdump"
78 - setenv SCRATCH_LOGDEV "device for scratch-fs external log"
79 - setenv SCRATCH_RTDEV "device for scratch-fs realtime data"
80 - setenv TEST_LOGDEV "device for test-fs external log"
81 - setenv TEST_RTDEV "device for test-fs realtime data"
82 - if TEST_LOGDEV and/or TEST_RTDEV, these will always be used.
83 - if SCRATCH_LOGDEV and/or SCRATCH_RTDEV, the USE_EXTERNAL
84 environment variable set to "yes" will enable their use.
85 - setenv DIFF_LENGTH "number of diff lines to print from a failed test",
86 by default 10, set to 0 to print the full diff
87 - setenv FSTYP "the filesystem you want to test", the filesystem
88 type is devised from the TEST_DEV device, but you may want to
89 override it; if unset, the default is 'xfs'
90 - setenv FSSTRESS_AVOID and/or FSX_AVOID, which contain options
91 added to the end of fsstresss and fsx invocations, respectively,
92 in case you wish to exclude certain operational modes from these
94 - set TEST_XFS_REPAIR_REBUILD=1 to have _check_xfs_filesystem
95 run xfs_repair -n to check the filesystem; xfs_repair to rebuild
96 metadata indexes; and xfs_repair -n (a third time) to check the
97 results of the rebuilding.
98 - set TEST_XFS_SCRUB=1 to have _check_xfs_filesystem run
99 xfs_scrub -vd to scrub the filesystem metadata online before
100 unmounting to run the offline check.
102 - or add a case to the switch in common/config assigning
103 these variables based on the hostname of your test
105 - or add these variables to a file called local.config and keep that
106 file in your workarea.
108 - if testing xfsdump, make sure the tape devices have a
109 tape which can be overwritten.
111 - make sure $TEST_DEV is a mounted XFS partition
112 - make sure that $SCRATCH_DEV or $SCRATCH_DEV_POOL contains nothing useful
117 - By default the tests suite will run xfs tests:
118 - ./check '*/001' '*/002' '*/003'
120 - Groups of tests maybe ran by: ./check -g [group(s)]
121 See the 'group' file for details on groups
122 - To randomize test order: ./check -r [test(s)]
123 - You can explicitly specify NFS/CIFS/OVERLAY, otherwise
124 the filesystem type will be autodetected from $TEST_DEV:
125 - for running nfs tests: ./check -nfs [test(s)]
126 - for running cifs/smb3 tests: ./check -cifs [test(s)]
127 - for overlay tests: ./check -overlay [test(s)]
128 The TEST and SCRATCH partitions should be pre-formatted
129 with another base fs, where the overlay dirs will be created
132 The check script tests the return value of each script, and
133 compares the output against the expected output. If the output
134 is not as expected, a diff will be output and an .out.bad file
135 will be produced for the failing test.
137 Unexpected console messages, crashes and hangs may be considered
138 to be failures but are not necessarily detected by the QA system.
140 __________________________
141 ADDING TO THE FSQA SUITE
142 __________________________
145 Creating new tests scripts:
147 Use the "new" script.
149 Test script environment:
151 When developing a new test script keep the following things in
152 mind. All of the environment variables and shell procedures are
153 available to the script once the "common/rc" file has been
156 1. The tests are run from an arbitrary directory. If you want to
157 do operations on an XFS filesystem (good idea, eh?), then do
158 one of the following:
160 (a) Create directories and files at will in the directory
161 $TEST_DIR ... this is within an XFS filesystem and world
162 writeable. You should cleanup when your test is done,
163 e.g. use a _cleanup shell procedure in the trap ... see
164 001 for an example. If you need to know, the $TEST_DIR
165 directory is within the filesystem on the block device
168 (b) mkfs a new XFS filesystem on $SCRATCH_DEV, and mount this
169 on $SCRATCH_MNT. Call the the _require_scratch function
170 on startup if you require use of the scratch partition.
171 _require_scratch does some checks on $SCRATCH_DEV &
172 $SCRATCH_MNT and makes sure they're unmounted. You should
173 cleanup when your test is done, and in particular unmount
175 Tests can make use of $SCRATCH_LOGDEV and $SCRATCH_RTDEV
176 for testing external log and realtime volumes - however,
177 these tests need to simply "pass" (e.g. cat $seq.out; exit
178 - or default to an internal log) in the common case where
179 these variables are not set.
181 2. You can safely create temporary files that are not part of the
182 filesystem tests (e.g. to catch output, prepare lists of things
183 to do, etc.) in files named $tmp.<anything>. The standard test
184 script framework created by "new" will initialize $tmp and
187 3. By default, tests are run as the same uid as the person
188 executing the control script "check" that runs the test scripts.
190 4. Some other useful shell procedures:
192 _get_fqdn - echo the host's fully qualified
195 _get_pids_by_name - one argument is a process name, and
196 return all of the matching pids on
199 _within_tolerance - fancy numerical "close enough is good
200 enough" filter for deterministic
201 output ... see comments in
202 common/filter for an explanation
204 _filter_date - turn ctime(3) format dates into the
205 string DATE for deterministic
208 _cat_passwd, - dump the content of the password
209 _cat_group or group file (both the local file
210 and the content of the NIS database
211 if it is likely to be present)
213 5. General recommendations, usage conventions, etc.:
214 - When the content of the password or group file is
215 required, get it using the _cat_passwd and _cat_group
216 functions, to ensure NIS information is included if NIS
218 - When calling getfacl in a test, pass the "-n" argument so
219 that numeric rather than symbolic identifiers are used in
221 - When creating a new test, it is possible to enter a custom name
222 for the file. Filenames are in form NNN-custom-name, where NNN
223 is automatically added by the ./new script as an unique ID,
224 and "custom-name" is the optional string entered into a prompt
225 in the ./new script. It can contain only alphanumeric characters
226 and dash. Note the "NNN-" part is added automatically.
230 Each test script has a name, e.g. 007, and an associated
231 verified output, e.g. 007.out.
233 It is important that the verified output is deterministic, and
234 part of the job of the test script is to filter the output to
235 make this so. Examples of the sort of things that need filtering:
242 - variable directory contents
243 - imprecise numbers, especially sizes and times
247 The script "check" may be used to run one or more tests.
249 Test number $seq is deemed to "pass" when:
250 (a) no "core" file is created,
251 (b) the file $seq.notrun is not created,
252 (c) the exit status is 0, and
253 (d) the output matches the verified output.
255 In the "not run" case (b), the $seq.notrun file should contain a
256 short one-line summary of why the test was not run. The standard
257 output is not checked, so this can be used for a more verbose
258 explanation and to provide feedback when the QA test is run
262 To force a non-zero exit status use:
268 won't have the desired effect because of the way the exit trap
271 The recent pass/fail history is maintained in the file "check.log".
272 The elapsed time for the most recent pass for each test is kept
275 The compare-failures script in tools/ may be used to compare failures
276 across multiple runs, given files containing stdout from those runs.
282 Send patches to the fstests mailing list at fstests@vger.kernel.org.