summaryrefslogtreecommitdiffstats
path: root/run-tests.sh
blob: 4e722d211067fd14f62ea8e1043f29d826ee8124 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
#!/bin/bash
# Copyright (c) 2013-2014 Red Hat, Inc. <http://www.redhat.com>
#

export TZ=UTC
force="no"
head="yes"
retry="yes"
tests=""
exit_on_failure="yes"
skip_bad_tests="yes"
skip_known_bugs="yes"
result_output="/tmp/gluster_regression.txt"
section_separator="========================================"
run_timeout=200
kill_after_time=5
# Option below preserves log tarballs for each run of a test separately
#       named: <test>-iteration-<n>.tar
# If set to any other value, then log tarball is just named after the test and
# overwritten in each iteration (saves space)
#       named: <test>.tar
# Use option -p to override default behavior
skip_preserve_logs="yes"

OSTYPE=$(uname -s)

# Function for use in generating filenames with increasing "-<n>" index
# In:
#       $1 basepath: Directory where file needs to be created
#       $2 filename: Name of the file sans extension
#       $3 extension: Extension string that would be appended to the generated
#               filename
# Out:
#       string of next available filename with appended "-<n>"
# Example:
#       Interested routines that want to create a file name, say foo-<n>.txt at
#       location /var/log/gluster would pass in "/var/log/gluster" "foo" "txt"
#       and be returned next available foo-<n> filename to create.
# Notes:
#       Function will not accept empty extension, and will return the same name
#       over and over (which can be fixed when there is a need for it)
function get_next_filename()
{
        local basepath=$1
        local filename=$2
        local extension=$3
        local next=1
        local tfilename="${filename}-${next}"
        while [ -e "${basepath}/${tfilename}.${extension}" ]; do
                next=$((next+1))
                tfilename="${filename}-${next}"
        done

        echo "$tfilename"
}

# Tar the gluster logs and generate a tarball named after the first parameter
# passed in to the function. Ideally the test name is passed to this function
# to generate the required tarball.
# Tarball name is further controlled by the variable skip_preserve_logs
function tar_logs()
{
        t=$1

        logdir=$(gluster --print-logdir)
        basetarname=$(basename "$t" .t)

        if [ -n "$logdir" ]
        then
                if [[ $skip_preserve_logs == "yes" ]]; then
                        savetarname=$(get_next_filename "${logdir}" \
                                "${basetarname}-iteration" "tar" \
                                | tail -1)
                else
                        savetarname="$basetarname"
                fi

                # Can't use --exclude here because NetBSD doesn't have it.
                # However, both it and Linux have -X to take patterns from
                # a file, so use that.
                (echo '*.tar'; echo .notar) > "${logdir}"/.notar \
                        && \
                tar -cf "${logdir}"/"${savetarname}".tar -X "${logdir}"/.notar \
                        "${logdir}"/* 2> /dev/null \
                        && \
                find "$logdir"/* -maxdepth 0 -name '*.tar' -prune \
                                        -o -exec rm -rf '{}' ';'

                echo "Logs preserved in tarball $savetarname.tar"
        else
                echo "Logs not preserved, as logdir is not set"
        fi
}

function check_dependencies()
{
    ## Check all dependencies are present
    MISSING=""

    # Check for dbench
    env dbench --usage > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING dbench"
    fi

    # Check for git
    env git --version > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING git"
    fi

    # Check for nfs-utils (Linux-only: built-in NetBSD with different name)
    if [ "x`uname -s`" = "xLinux" ] ; then
      env mount.nfs -V > /dev/null 2>&1
      if [ $? -ne 0 ]; then
          MISSING="$MISSING nfs-utils"
      fi
    fi

    # Check for netstat
    env netstat --version > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING netstat"
    fi

    # Check for the Perl Test Harness
    env prove --version > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING perl-Test-Harness"
    fi

    which json_verify > /dev/null
    if [ $? -ne 0 ]; then
        MISSING="$MISSING json_verify"
    fi

    # Check for XFS programs (Linux Only: NetBSD does without)
    if [ "x`uname -s`" = "xLinux" ] ; then
      env mkfs.xfs -V > /dev/null 2>&1
      if [ $? -ne 0 ]; then
          MISSING="$MISSING xfsprogs"
      fi
    fi

    # Check for attr
    env getfattr --version > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING attr"
    fi

    # Check for pidof
    pidof pidof > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING pidof"
    fi

    # Check for netstat
    env netstat --version > /dev/null 2>&1
    if [ $? -ne 0 ]; then
        MISSING="$MISSING netstat"
    fi

    # check for psutil python package
    test `uname -s` == "Darwin" || test `uname -s` == "FreeBSD" && {
        pip show psutil | grep -q psutil >/dev/null 2>&1
        if [ $? -ne 0 ]; then
            MISSING="$MISSING psutil"
        fi
    }

    ## If dependencies are missing, warn the user and abort
    if [ "x$MISSING" != "x" ]; then
        test "x${force}" != "xyes" && echo "Aborting."
        echo
        echo "The following required tools are missing:"
        echo
        for pkg in $MISSING; do
            echo "  * $pkg"
        done
        echo
        test "x${force}" = "xyes" && return
        echo "Please install them and try again."
        echo
        exit 2
    fi
}

function check_location()
{
    regression_testsdir=$(dirname $0);

    if [ ! -f ${regression_testsdir}/tests/include.rc ]; then
        echo "Aborting."
        echo
        echo "The tests/ subdirectory seems to be missing."
        echo
        echo "Please correct the problem and try again."
        echo
        exit 1
    fi
}

function check_user()
{
    # If we're not running as root, warn the user and abort
    MYUID=`/usr/bin/id -u`
    if [ 0${MYUID} -ne 0 ]; then
        echo "Aborting."
        echo
        echo "The GlusterFS Test Framework must be run as root."
        echo
        echo "Please change to the root user and try again."
        echo
        exit 3
    fi
}

function match()
{
        # Patterns considered valid:
        # 0. Empty means everything
        #    ""           matches ** i.e all
        # 1. full or partial file/directory names
        #   basic         matches tests/basic
        #   basic/afr     matches tests/basic/afr
        # 2. globs
        #   basic/*       matches all files and directories in basic
        #   basic/*/      matches subdirectories in basic (afr|ec)
        # 3. numbered bug matching
        #   884455        matches bugs/bug-884455.t
        #   859927        matches bugs/859927, bugs/bug-859927.t
        #   1015990       matches /bugs/bug-1015990-rep.t, bug-1015990.t
        # ...lots of other cases accepted as well, since globbing is tricky.
        local t=$1
        shift
        local a
        local match=1
        if [ -z "$@" ]; then
            match=0
            return $match
        fi
        for a in $@ ; do
            case "$t" in
                *$a*)
                    match=0
                    ;;
            esac
        done
        return $match
}

# Tests can have comment lines with some comma separated values within them.
# Key names used to determine test status are
# G_TESTDEF_TEST_STATUS_CENTOS6
# G_TESTDEF_TEST_STATUS_NETBSD7
# Some examples:
# G_TESTDEF_TEST_STATUS_CENTOS6=BAD_TEST,BUG=123456
# G_TESTDEF_TEST_STATUS_NETBSD7=KNOWN_ISSUE,BUG=4444444
# G_TESTDEF_TEST_STATUS_CENTOS6=BAD_TEST,BUG=123456;555555
# You can change status of test to enabled or delete the line only if all the
# bugs are closed or modified or if the patch fixes it.
function get_test_status ()
{
    local test_name=$1
    local host_os=""
    local result=""

    host_os=$(uname -s)

    case "$host_os" in
    # Leaving out the logic to determine the particular distro and version
    # for later. Why does the key have the distro and version then?
    # Because changing the key in all test files would be very big process
    # updating just this function with a better logic much simpler.
    Linux)
        result=$(grep -e "^#G_TESTDEF_TEST_STATUS_CENTOS6" $test_name | \
                 awk -F"," {'print $1'} | awk -F"=" {'print $2'}) ;;
    NetBSD)
        result=$(grep -e "^#G_TESTDEF_TEST_STATUS_NETBSD7" $test_name | \
                 awk -F"," {'print $1'} | awk -F"=" {'print $2'}) ;;
    *)
        result="ENABLED" ;;
    esac

    echo "$result"

}

function get_bug_list_for_disabled_test ()
{
    local test_name=$1
    local host_os=""
    local result=""

    host_os=$(uname -s)

    case "$host_os" in
    # Leaving out the logic to determine the particular distro and version
    # for later. Why does the key have the distro and version then?
    # Because changing the key in all test files would be very big process
    # updating just this function with a better logic much simpler.
    Linux)
        result=$(grep -e "^#G_TESTDEF_TEST_STATUS_CENTOS6" $test_name | \
                 awk -F"," {'print $2'} | awk -F"=" {'print $2'}) ;;
    NetBSD)
        result=$(grep -e "^#G_TESTDEF_TEST_STATUS_NETBSD7" $test_name | \
                 awk -F"," {'print $2'} | awk -F"=" {'print $2'}) ;;
    *)
        result="0000000" ;;
    esac

    echo "$result"

}

function run_tests()
{
    RES=0
    FAILED=''
    TESTS_NEEDED_RETRY=''
    GENERATED_CORE=''
    total_tests=0
    selected_tests=0
    skipped_bad_tests=0
    skipped_known_issue_tests=0
    total_run_tests=0

    # key = path of .t file; value = time taken to run the .t file
    declare -A ELAPSEDTIMEMAP

    # Test if -k is supported for timeout command
    # This is not supported on centos6, but spuported on centos7
    # The flags is required for running the command in both flavors
    timeout_cmd_exists="yes"
    timeout -k 1 10 echo "testing 'timeout' command"
    if [ $? -ne 0 ]; then
        timeout_cmd_exists="no"
    fi

    for t in $(find ${regression_testsdir}/tests -name '*.t' \
               | LC_COLLATE=C sort) ; do
        old_cores=$(ls /*-*.core 2> /dev/null | wc -l)
        total_tests=$((total_tests+1))
        if match $t "$@" ; then
            selected_tests=$((selected_tests+1))
            echo
            echo $section_separator$section_separator
            if [[ $(get_test_status $t) == "BAD_TEST" ]] && \
               [[ $skip_bad_tests == "yes" ]]
            then
                skipped_bad_tests=$((skipped_bad_tests+1))
                echo "Skipping bad test file $t"
                echo "Reason: bug(s):" $(get_bug_list_for_disabled_test $t)
                echo $section_separator$section_separator
                echo
                continue
            fi
            if [[ $(get_test_status $t) == "KNOWN_ISSUE" ]] && \
               [[ $skip_known_bugs == "yes" ]]
            then
                skipped_known_issue_tests=$((skipped_known_issue_tests+1))
                echo "Skipping test file $t due to known issue"
                echo "Reason: bug(s):" $(get_bug_list_for_disabled_test $t)
                echo $section_separator$section_separator
                echo
                continue
            fi
            total_run_tests=$((total_run_tests+1))
            echo "[$(date +%H:%M:%S)] Running tests in file $t"
            starttime="$(date +%s)"

            local cmd_timeout=$run_timeout;
            if [ ${timeout_cmd_exists} == "yes" ]; then
                if [ $(grep -c "SCRIPT_TIMEOUT=" ${t}) == 1 ] ; then
                    cmd_timeout=$(grep "SCRIPT_TIMEOUT=" ${t} | cut -f2 -d'=');
                    echo "Timeout set is ${cmd_timeout}, default ${run_timeout}"
                fi
                timeout -k ${kill_after_time} ${cmd_timeout} prove -vmfe '/bin/bash' ${t}
            else
                prove -vmfe '/bin/bash' ${t}
            fi
            TMP_RES=$?
            ELAPSEDTIMEMAP[$t]=`expr $(date +%s) - $starttime`
            tar_logs "$t"

            # timeout always return 124 if it is actually a timeout.
            if ((${TMP_RES} == 124)); then
                echo "${t} timed out after ${cmd_timeout} seconds"
            fi

            if [ ${TMP_RES} -ne 0 ]  && [ "x${retry}" = "xyes" ] ; then
                echo "$t: bad status $TMP_RES"
                echo ""
                echo "       *********************************"
                echo "       *       REGRESSION FAILED       *"
                echo "       * Retrying failed tests in case *"
                echo "       * we got some spurious failures *"
                echo "       *********************************"
                echo ""

                if [ ${timeout_cmd_exists} == "yes" ]; then
                    timeout -k ${kill_after_time} ${cmd_timeout} prove -vmfe '/bin/bash' ${t}
                else
                    prove -vmfe '/bin/bash' ${t}
                fi
                TMP_RES=$?
                tar_logs "$t"

                if ((${TMP_RES} == 124)); then
                    echo "${t} timed out after ${cmd_timeout} seconds"
                fi

                TESTS_NEEDED_RETRY="${TESTS_NEEDED_RETRY}${t} "
            fi
            if [ ${TMP_RES} -ne 0 ] ; then
                RES=${TMP_RES}
                FAILED="${FAILED}${t} "
            fi

            new_cores=$(ls /*-*.core 2> /dev/null | wc -l)
            if [ x"$new_cores" != x"$old_cores" ]; then
                core_diff=$((new_cores-old_cores))
                echo "$t: $core_diff new core files"
                RES=1
                GENERATED_CORE="${GENERATED_CORE}${t} "
            fi
            echo "End of test $t"
            echo $section_separator$section_separator
            echo
            if [ $RES -ne 0 ] && [ x"$exit_on_failure" = "xyes" ] ; then
                break;
            fi
        fi
    done
    echo
    echo "Run complete"
    echo $section_separator$section_separator
    echo "Number of tests found:                             $total_tests"
    echo "Number of tests selected for run based on pattern: $selected_tests"
    echo "Number of tests skipped as they were marked bad:   $skipped_bad_tests"
    echo "Number of tests skipped because of known_issues:   $skipped_known_issue_tests"
    echo "Number of tests that were run:                     $total_run_tests"

    echo
    echo "Tests ordered by time taken, slowest to fastest: "
    echo $section_separator$section_separator
    for key in "${!ELAPSEDTIMEMAP[@]}"
    do
        echo "$key  -  ${ELAPSEDTIMEMAP["$key"]} second"
    done | sort -rn -k3

    # Output the errors into a file
    echo > "${result_output}"
    if [ ${RES} -ne 0 ] ; then
        FAILED=$( echo ${FAILED} | tr ' ' '\n' | sort -u )
        FAILED_COUNT=$( echo -n "${FAILED}" | grep -c '^' )
        echo -e "\n$FAILED_COUNT test(s) failed \n${FAILED}" >> "${result_output}"
        GENERATED_CORE=$( echo  ${GENERATED_CORE} | tr ' ' '\n' | sort -u )
        GENERATED_CORE_COUNT=$( echo -n "${GENERATED_CORE}" | grep -c '^' )
        echo -e "\n$GENERATED_CORE_COUNT test(s) generated core \n${GENERATED_CORE}" >> "${result_output}"
        cat "${result_output}"
    fi
    TESTS_NEEDED_RETRY=$( echo ${TESTS_NEEDED_RETRY} | tr ' ' '\n' | sort -u )
    RETRY_COUNT=$( echo -n "${TESTS_NEEDED_RETRY}" | grep -c '^' )
    if [ ${RETRY_COUNT} -ne 0 ] ; then
        echo -e "\n${RETRY_COUNT} test(s) needed retry \n${TESTS_NEEDED_RETRY}"
    fi

    echo
    echo "Result is $RES"
    echo
    return ${RES}
}

function run_head_tests()
{
    [ -d ${regression_testsdir}/.git ] || return 0

    # The git command needs $cwd to be within the repository, but run_tests
    # needs it to be back where we started.
    pushd $regression_testsdir
    git_cmd="git diff-tree --no-commit-id --name-only --diff-filter=ACMRTUXB"
    htests=$($git_cmd -r HEAD tests | grep '.t$')
    popd
    [ -n "$htests" ] || return 0

    # Perhaps it's not ideal that we'll end up re-running these tests, but the
    # gains from letting them fail fast in the first round should outweigh the
    # losses from running them again in the second.  OTOH, with so many of our
    # tests being non-deterministic, maybe it doesn't hurt to give the newest
    # tests an extra workout.
    run_tests "$htests"
}

function parse_args () {
    args=`getopt frcbkphHo:t: "$@"`
    set -- $args
    while [ $# -gt 0 ]; do
        case "$1" in
        -f)    force="yes" ;;
        -h)    head="no" ;;
        -H)    head="only" ;;
        -r)    retry="yes" ;;
        -c)    exit_on_failure="no" ;;
        -b)    skip_bad_tests="no" ;;
        -k)    skip_known_bugs="no" ;;
        -p)    skip_preserve_logs="no" ;;
        -o)    result_output="$2"; shift;;
        -t)    run_timeout="$2"; shift;;
        --)    shift; break;;
        esac
        shift
    done
    tests="$@"
}


echo
echo ... GlusterFS Test Framework ...
echo

# Get user options
parse_args "$@"

# Make sure we're running as the root user
check_user

# Make sure the needed programs are available
check_dependencies

# Check we're running from the right location
check_location

# Run the tests
if [ x"$head" != x"no" ]; then
        run_head_tests || exit 1
fi
if [ x"$head" != x"only" ]; then
        run_tests "$tests" || exit 1
fi