]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
1 | #!/bin/bash -x |
2 | # | |
3 | # Copyright (C) 2014 Red Hat <contact@redhat.com> | |
4 | # | |
5 | # Author: Loic Dachary <loic@dachary.org> | |
6 | # | |
7 | # This program is free software; you can redistribute it and/or modify | |
8 | # it under the terms of the GNU Library Public License as published by | |
9 | # the Free Software Foundation; either version 2, or (at your option) | |
10 | # any later version. | |
11 | # | |
12 | # This program is distributed in the hope that it will be useful, | |
13 | # but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | # GNU Library Public License for more details. | |
16 | # | |
17 | source $(dirname $0)/../detect-build-env-vars.sh | |
18 | source $CEPH_ROOT/qa/workunits/ceph-helpers.sh | |
19 | ||
20 | # Test development and debugging | |
21 | # Set to "yes" in order to ignore diff errors and save results to update test | |
22 | getjson="no" | |
23 | ||
24 | termwidth=$(stty -a | head -1 | sed -e 's/.*columns \([0-9]*\).*/\1/') | |
25 | if test -n "$termwidth" -a "$termwidth" != "0"; then termwidth="-W ${termwidth}"; fi | |
26 | ||
27 | # Ignore the epoch and filter out the attr '_' value because it has date information and won't match | |
28 | jqfilter='.inconsistents | (.[].shards[].attrs[] | select(.name == "_") | .value) |= "----Stripped-by-test----"' | |
29 | sortkeys='import json; import sys ; JSON=sys.stdin.read() ; ud = json.loads(JSON) ; print json.dumps(ud, sort_keys=True, indent=2)' | |
30 | ||
31 | # Remove items are not consistent across runs, the pg interval and client | |
32 | sedfilter='s/\([ ]*\"\(selected_\)*object_info\":.*head[(]\)[^[:space:]]* [^[:space:]]* \(.*\)/\1\3/' | |
33 | ||
34 | function run() { | |
35 | local dir=$1 | |
36 | shift | |
37 | ||
38 | export CEPH_MON="127.0.0.1:7107" # git grep '\<7107\>' : there must be only one | |
39 | export CEPH_ARGS | |
40 | CEPH_ARGS+="--fsid=$(uuidgen) --auth-supported=none " | |
41 | CEPH_ARGS+="--mon-host=$CEPH_MON " | |
42 | ||
43 | local funcs=${@:-$(set | sed -n -e 's/^\(TEST_[0-9a-z_]*\) .*/\1/p')} | |
44 | for func in $funcs ; do | |
45 | $func $dir || return 1 | |
46 | done | |
47 | } | |
48 | ||
49 | function add_something() { | |
50 | local dir=$1 | |
51 | local poolname=$2 | |
52 | local obj=${3:-SOMETHING} | |
53 | local scrub=${4:-noscrub} | |
54 | ||
55 | if [ "$scrub" = "noscrub" ]; | |
56 | then | |
57 | ceph osd set noscrub || return 1 | |
58 | ceph osd set nodeep-scrub || return 1 | |
59 | else | |
60 | ceph osd unset noscrub || return 1 | |
61 | ceph osd unset nodeep-scrub || return 1 | |
62 | fi | |
63 | ||
64 | local payload=ABCDEF | |
65 | echo $payload > $dir/ORIGINAL | |
66 | rados --pool $poolname put $obj $dir/ORIGINAL || return 1 | |
67 | } | |
68 | ||
69 | # | |
70 | # Corrupt one copy of a replicated pool | |
71 | # | |
72 | function TEST_corrupt_and_repair_replicated() { | |
73 | local dir=$1 | |
74 | local poolname=rbd | |
75 | ||
76 | setup $dir || return 1 | |
77 | run_mon $dir a --osd_pool_default_size=2 || return 1 | |
78 | run_mgr $dir x || return 1 | |
79 | run_osd $dir 0 || return 1 | |
80 | run_osd $dir 1 || return 1 | |
81 | wait_for_clean || return 1 | |
82 | ||
83 | add_something $dir $poolname || return 1 | |
84 | corrupt_and_repair_one $dir $poolname $(get_not_primary $poolname SOMETHING) || return 1 | |
85 | # Reproduces http://tracker.ceph.com/issues/8914 | |
86 | corrupt_and_repair_one $dir $poolname $(get_primary $poolname SOMETHING) || return 1 | |
87 | ||
88 | teardown $dir || return 1 | |
89 | } | |
90 | ||
91 | function corrupt_and_repair_two() { | |
92 | local dir=$1 | |
93 | local poolname=$2 | |
94 | local first=$3 | |
95 | local second=$4 | |
96 | ||
97 | # | |
98 | # 1) remove the corresponding file from the OSDs | |
99 | # | |
100 | pids="" | |
101 | run_in_background pids objectstore_tool $dir $first SOMETHING remove | |
102 | run_in_background pids objectstore_tool $dir $second SOMETHING remove | |
103 | wait_background pids | |
104 | return_code=$? | |
105 | if [ $return_code -ne 0 ]; then return $return_code; fi | |
106 | ||
107 | # | |
108 | # 2) repair the PG | |
109 | # | |
110 | local pg=$(get_pg $poolname SOMETHING) | |
111 | repair $pg | |
112 | # | |
113 | # 3) The files must be back | |
114 | # | |
115 | pids="" | |
116 | run_in_background pids objectstore_tool $dir $first SOMETHING list-attrs | |
117 | run_in_background pids objectstore_tool $dir $second SOMETHING list-attrs | |
118 | wait_background pids | |
119 | return_code=$? | |
120 | if [ $return_code -ne 0 ]; then return $return_code; fi | |
121 | ||
122 | rados --pool $poolname get SOMETHING $dir/COPY || return 1 | |
123 | diff $dir/ORIGINAL $dir/COPY || return 1 | |
124 | } | |
125 | ||
126 | # | |
127 | # 1) add an object | |
128 | # 2) remove the corresponding file from a designated OSD | |
129 | # 3) repair the PG | |
130 | # 4) check that the file has been restored in the designated OSD | |
131 | # | |
132 | function corrupt_and_repair_one() { | |
133 | local dir=$1 | |
134 | local poolname=$2 | |
135 | local osd=$3 | |
136 | ||
137 | # | |
138 | # 1) remove the corresponding file from the OSD | |
139 | # | |
140 | objectstore_tool $dir $osd SOMETHING remove || return 1 | |
141 | # | |
142 | # 2) repair the PG | |
143 | # | |
144 | local pg=$(get_pg $poolname SOMETHING) | |
145 | repair $pg | |
146 | # | |
147 | # 3) The file must be back | |
148 | # | |
149 | objectstore_tool $dir $osd SOMETHING list-attrs || return 1 | |
150 | rados --pool $poolname get SOMETHING $dir/COPY || return 1 | |
151 | diff $dir/ORIGINAL $dir/COPY || return 1 | |
152 | } | |
153 | ||
154 | function corrupt_and_repair_erasure_coded() { | |
155 | local dir=$1 | |
156 | local poolname=$2 | |
157 | ||
158 | add_something $dir $poolname || return 1 | |
159 | ||
160 | local primary=$(get_primary $poolname SOMETHING) | |
161 | local -a osds=($(get_osds $poolname SOMETHING | sed -e "s/$primary//")) | |
162 | local not_primary_first=${osds[0]} | |
163 | local not_primary_second=${osds[1]} | |
164 | ||
165 | # Reproduces http://tracker.ceph.com/issues/10017 | |
166 | corrupt_and_repair_one $dir $poolname $primary || return 1 | |
167 | # Reproduces http://tracker.ceph.com/issues/10409 | |
168 | corrupt_and_repair_one $dir $poolname $not_primary_first || return 1 | |
169 | corrupt_and_repair_two $dir $poolname $not_primary_first $not_primary_second || return 1 | |
170 | corrupt_and_repair_two $dir $poolname $primary $not_primary_first || return 1 | |
171 | ||
172 | } | |
173 | ||
174 | function create_ec_pool() { | |
175 | local pool_name=$1 | |
176 | local allow_overwrites=$2 | |
177 | ||
178 | ceph osd erasure-code-profile set myprofile ruleset-failure-domain=osd $3 $4 $5 $6 $7 || return 1 | |
179 | ||
180 | ceph osd pool create "$poolname" 1 1 erasure myprofile || return 1 | |
181 | ||
182 | if [ "$allow_overwrites" = "true" ]; then | |
183 | ceph osd pool set "$poolname" allow_ec_overwrites true || return 1 | |
184 | fi | |
185 | ||
186 | wait_for_clean || return 1 | |
187 | return 0 | |
188 | } | |
189 | ||
190 | function auto_repair_erasure_coded() { | |
191 | local dir=$1 | |
192 | local allow_overwrites=$2 | |
193 | local poolname=ecpool | |
194 | ||
195 | # Launch a cluster with 5 seconds scrub interval | |
196 | setup $dir || return 1 | |
197 | run_mon $dir a || return 1 | |
198 | run_mgr $dir x || return 1 | |
199 | local ceph_osd_args="--osd-scrub-auto-repair=true \ | |
200 | --osd-deep-scrub-interval=5 \ | |
201 | --osd-scrub-max-interval=5 \ | |
202 | --osd-scrub-min-interval=5 \ | |
203 | --osd-scrub-interval-randomize-ratio=0" | |
204 | for id in $(seq 0 2) ; do | |
205 | if [ "$allow_overwrites" = "true" ]; then | |
206 | run_osd_bluestore $dir $id $ceph_osd_args || return 1 | |
207 | else | |
208 | run_osd $dir $id $ceph_osd_args || return 1 | |
209 | fi | |
210 | done | |
211 | wait_for_clean || return 1 | |
212 | ||
213 | # Create an EC pool | |
214 | create_ec_pool $poolname $allow_overwrites k=2 m=1 || return 1 | |
215 | ||
216 | # Put an object | |
217 | local payload=ABCDEF | |
218 | echo $payload > $dir/ORIGINAL | |
219 | rados --pool $poolname put SOMETHING $dir/ORIGINAL || return 1 | |
220 | ||
221 | # Remove the object from one shard physically | |
222 | # Restarted osd get $ceph_osd_args passed | |
223 | objectstore_tool $dir $(get_not_primary $poolname SOMETHING) SOMETHING remove || return 1 | |
224 | # Wait for auto repair | |
225 | local pgid=$(get_pg $poolname SOMETHING) | |
226 | wait_for_scrub $pgid "$(get_last_scrub_stamp $pgid)" | |
227 | wait_for_clean || return 1 | |
228 | # Verify - the file should be back | |
229 | # Restarted osd get $ceph_osd_args passed | |
230 | objectstore_tool $dir $(get_not_primary $poolname SOMETHING) SOMETHING list-attrs || return 1 | |
231 | rados --pool $poolname get SOMETHING $dir/COPY || return 1 | |
232 | diff $dir/ORIGINAL $dir/COPY || return 1 | |
233 | ||
234 | # Tear down | |
235 | teardown $dir || return 1 | |
236 | } | |
237 | ||
238 | function TEST_auto_repair_erasure_coded_appends() { | |
239 | auto_repair_erasure_coded $1 false | |
240 | } | |
241 | ||
242 | function TEST_auto_repair_erasure_coded_overwrites() { | |
243 | auto_repair_erasure_coded $1 true | |
244 | } | |
245 | ||
246 | function corrupt_and_repair_jerasure() { | |
247 | local dir=$1 | |
248 | local allow_overwrites=$2 | |
249 | local poolname=ecpool | |
250 | ||
251 | setup $dir || return 1 | |
252 | run_mon $dir a || return 1 | |
253 | run_mgr $dir x || return 1 | |
254 | for id in $(seq 0 3) ; do | |
255 | if [ "$allow_overwrites" = "true" ]; then | |
256 | run_osd_bluestore $dir $id || return 1 | |
257 | else | |
258 | run_osd $dir $id || return 1 | |
259 | fi | |
260 | done | |
261 | wait_for_clean || return 1 | |
262 | ||
263 | create_ec_pool $poolname $allow_overwrites k=2 m=2 || return 1 | |
264 | corrupt_and_repair_erasure_coded $dir $poolname || return 1 | |
265 | ||
266 | teardown $dir || return 1 | |
267 | } | |
268 | ||
269 | function TEST_corrupt_and_repair_jerasure_appends() { | |
270 | corrupt_and_repair_jerasure $1 | |
271 | } | |
272 | ||
273 | function TEST_corrupt_and_repair_jerasure_overwrites() { | |
274 | corrupt_and_repair_jerasure $1 true | |
275 | } | |
276 | ||
277 | function corrupt_and_repair_lrc() { | |
278 | local dir=$1 | |
279 | local allow_overwrites=$2 | |
280 | local poolname=ecpool | |
281 | ||
282 | setup $dir || return 1 | |
283 | run_mon $dir a || return 1 | |
284 | run_mgr $dir x || return 1 | |
285 | for id in $(seq 0 9) ; do | |
286 | if [ "$allow_overwrites" = "true" ]; then | |
287 | run_osd_bluestore $dir $id || return 1 | |
288 | else | |
289 | run_osd $dir $id || return 1 | |
290 | fi | |
291 | done | |
292 | wait_for_clean || return 1 | |
293 | ||
294 | create_ec_pool $poolname $allow_overwrites k=4 m=2 l=3 plugin=lrc || return 1 | |
295 | corrupt_and_repair_erasure_coded $dir $poolname || return 1 | |
296 | ||
297 | teardown $dir || return 1 | |
298 | } | |
299 | ||
300 | function TEST_corrupt_and_repair_lrc_appends() { | |
301 | corrupt_and_repair_jerasure $1 | |
302 | } | |
303 | ||
304 | function TEST_corrupt_and_repair_lrc_overwrites() { | |
305 | corrupt_and_repair_jerasure $1 true | |
306 | } | |
307 | ||
308 | function unfound_erasure_coded() { | |
309 | local dir=$1 | |
310 | local allow_overwrites=$2 | |
311 | local poolname=ecpool | |
312 | local payload=ABCDEF | |
313 | ||
314 | setup $dir || return 1 | |
315 | run_mon $dir a || return 1 | |
316 | run_mgr $dir x || return 1 | |
317 | for id in $(seq 0 3) ; do | |
318 | if [ "$allow_overwrites" = "true" ]; then | |
319 | run_osd_bluestore $dir $id || return 1 | |
320 | else | |
321 | run_osd $dir $id || return 1 | |
322 | fi | |
323 | done | |
324 | wait_for_clean || return 1 | |
325 | ||
326 | create_ec_pool $poolname $allow_overwrites k=2 m=2 || return 1 | |
327 | ||
328 | add_something $dir $poolname || return 1 | |
329 | ||
330 | local primary=$(get_primary $poolname SOMETHING) | |
331 | local -a osds=($(get_osds $poolname SOMETHING | sed -e "s/$primary//")) | |
332 | local not_primary_first=${osds[0]} | |
333 | local not_primary_second=${osds[1]} | |
334 | local not_primary_third=${osds[2]} | |
335 | ||
336 | # | |
337 | # 1) remove the corresponding file from the OSDs | |
338 | # | |
339 | pids="" | |
340 | run_in_background pids objectstore_tool $dir $not_primary_first SOMETHING remove | |
341 | run_in_background pids objectstore_tool $dir $not_primary_second SOMETHING remove | |
342 | run_in_background pids objectstore_tool $dir $not_primary_third SOMETHING remove | |
343 | wait_background pids | |
344 | return_code=$? | |
345 | if [ $return_code -ne 0 ]; then return $return_code; fi | |
346 | ||
347 | # | |
348 | # 2) repair the PG | |
349 | # | |
350 | local pg=$(get_pg $poolname SOMETHING) | |
351 | repair $pg | |
352 | # | |
353 | # 3) check pg state | |
354 | # | |
355 | # it may take a bit to appear due to mon/mgr asynchrony | |
356 | for f in `seq 1 60`; do | |
357 | ceph -s | grep "1/1 unfound" && break | |
358 | sleep 1 | |
359 | done | |
360 | ceph -s|grep "4 osds: 4 up, 4 in" || return 1 | |
361 | ceph -s|grep "1/1 unfound" || return 1 | |
362 | ||
363 | teardown $dir || return 1 | |
364 | } | |
365 | ||
366 | function TEST_unfound_erasure_coded_appends() { | |
367 | unfound_erasure_coded $1 | |
368 | } | |
369 | ||
370 | function TEST_unfound_erasure_coded_overwrites() { | |
371 | unfound_erasure_coded $1 true | |
372 | } | |
373 | ||
374 | # | |
375 | # list_missing for EC pool | |
376 | # | |
377 | function list_missing_erasure_coded() { | |
378 | local dir=$1 | |
379 | local allow_overwrites=$2 | |
380 | local poolname=ecpool | |
381 | ||
382 | setup $dir || return 1 | |
383 | run_mon $dir a || return 1 | |
384 | run_mgr $dir x || return 1 | |
385 | for id in $(seq 0 2) ; do | |
386 | if [ "$allow_overwrites" = "true" ]; then | |
387 | run_osd_bluestore $dir $id || return 1 | |
388 | else | |
389 | run_osd $dir $id || return 1 | |
390 | fi | |
391 | done | |
392 | wait_for_clean || return 1 | |
393 | ||
394 | create_ec_pool $poolname $allow_overwrites k=2 m=1 || return 1 | |
395 | ||
396 | # Put an object and remove the two shards (including primary) | |
397 | add_something $dir $poolname MOBJ0 || return 1 | |
398 | local -a osds0=($(get_osds $poolname MOBJ0)) | |
399 | ||
400 | # Put another object and remove two shards (excluding primary) | |
401 | add_something $dir $poolname MOBJ1 || return 1 | |
402 | local -a osds1=($(get_osds $poolname MOBJ1)) | |
403 | ||
404 | # Stop all osd daemons | |
405 | for id in $(seq 0 2) ; do | |
406 | kill_daemons $dir TERM osd.$id >&2 < /dev/null || return 1 | |
407 | done | |
408 | ||
409 | id=${osds0[0]} | |
410 | ceph-objectstore-tool --data-path $dir/$id --enable-experimental-unrecoverable-data-corrupting-features=bluestore \ | |
411 | MOBJ0 remove || return 1 | |
412 | id=${osds0[1]} | |
413 | ceph-objectstore-tool --data-path $dir/$id --enable-experimental-unrecoverable-data-corrupting-features=bluestore \ | |
414 | MOBJ0 remove || return 1 | |
415 | ||
416 | id=${osds1[1]} | |
417 | ceph-objectstore-tool --data-path $dir/$id --enable-experimental-unrecoverable-data-corrupting-features=bluestore \ | |
418 | MOBJ1 remove || return 1 | |
419 | id=${osds1[2]} | |
420 | ceph-objectstore-tool --data-path $dir/$id --enable-experimental-unrecoverable-data-corrupting-features=bluestore \ | |
421 | MOBJ1 remove || return 1 | |
422 | ||
423 | for id in $(seq 0 2) ; do | |
424 | activate_osd $dir $id >&2 || return 1 | |
425 | done | |
426 | wait_for_clean || return 1 | |
427 | ||
428 | # Get get - both objects should in the same PG | |
429 | local pg=$(get_pg $poolname MOBJ0) | |
430 | ||
431 | # Repair the PG, which triggers the recovering, | |
432 | # and should mark the object as unfound | |
433 | repair $pg | |
434 | ||
435 | for i in $(seq 0 120) ; do | |
436 | [ $i -lt 60 ] || return 1 | |
437 | matches=$(ceph pg $pg list_missing | egrep "MOBJ0|MOBJ1" | wc -l) | |
438 | [ $matches -eq 2 ] && break | |
439 | done | |
440 | ||
441 | teardown $dir || return 1 | |
442 | } | |
443 | ||
444 | function TEST_list_missing_erasure_coded_appends() { | |
445 | list_missing_erasure_coded $1 false | |
446 | } | |
447 | ||
448 | function TEST_list_missing_erasure_coded_overwrites() { | |
449 | list_missing_erasure_coded $1 true | |
450 | } | |
451 | ||
452 | # | |
453 | # Corrupt one copy of a replicated pool | |
454 | # | |
455 | function TEST_corrupt_scrub_replicated() { | |
456 | local dir=$1 | |
457 | local poolname=csr_pool | |
458 | local total_objs=15 | |
459 | ||
460 | setup $dir || return 1 | |
461 | run_mon $dir a --osd_pool_default_size=2 || return 1 | |
462 | run_mgr $dir x || return 1 | |
463 | run_osd $dir 0 || return 1 | |
464 | run_osd $dir 1 || return 1 | |
465 | wait_for_clean || return 1 | |
466 | ||
467 | ceph osd pool create $poolname 1 1 || return 1 | |
468 | wait_for_clean || return 1 | |
469 | ||
470 | for i in $(seq 1 $total_objs) ; do | |
471 | objname=ROBJ${i} | |
472 | add_something $dir $poolname $objname || return 1 | |
473 | ||
474 | rados --pool $poolname setomapheader $objname hdr-$objname || return 1 | |
475 | rados --pool $poolname setomapval $objname key-$objname val-$objname || return 1 | |
476 | ||
477 | # Alternate corruption between osd.0 and osd.1 | |
478 | local osd=$(expr $i % 2) | |
479 | ||
480 | case $i in | |
481 | 1) | |
482 | # Size (deep scrub data_digest too) | |
483 | local payload=UVWXYZZZ | |
484 | echo $payload > $dir/CORRUPT | |
485 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
486 | ;; | |
487 | ||
488 | 2) | |
489 | # digest (deep scrub only) | |
490 | local payload=UVWXYZ | |
491 | echo $payload > $dir/CORRUPT | |
492 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
493 | ;; | |
494 | ||
495 | 3) | |
496 | # missing | |
497 | objectstore_tool $dir $osd $objname remove || return 1 | |
498 | ;; | |
499 | ||
500 | 4) | |
501 | # Modify omap value (deep scrub only) | |
502 | objectstore_tool $dir $osd $objname set-omap key-$objname $dir/CORRUPT || return 1 | |
503 | ;; | |
504 | ||
505 | 5) | |
506 | # Delete omap key (deep scrub only) | |
507 | objectstore_tool $dir $osd $objname rm-omap key-$objname || return 1 | |
508 | ;; | |
509 | ||
510 | 6) | |
511 | # Add extra omap key (deep scrub only) | |
512 | echo extra > $dir/extra-val | |
513 | objectstore_tool $dir $osd $objname set-omap key2-$objname $dir/extra-val || return 1 | |
514 | rm $dir/extra-val | |
515 | ;; | |
516 | ||
517 | 7) | |
518 | # Modify omap header (deep scrub only) | |
519 | echo -n newheader > $dir/hdr | |
520 | objectstore_tool $dir $osd $objname set-omaphdr $dir/hdr || return 1 | |
521 | rm $dir/hdr | |
522 | ;; | |
523 | ||
524 | 8) | |
525 | rados --pool $poolname setxattr $objname key1-$objname val1-$objname || return 1 | |
526 | rados --pool $poolname setxattr $objname key2-$objname val2-$objname || return 1 | |
527 | ||
528 | # Break xattrs | |
529 | echo -n bad-val > $dir/bad-val | |
530 | objectstore_tool $dir $osd $objname set-attr _key1-$objname $dir/bad-val || return 1 | |
531 | objectstore_tool $dir $osd $objname rm-attr _key2-$objname || return 1 | |
532 | echo -n val3-$objname > $dir/newval | |
533 | objectstore_tool $dir $osd $objname set-attr _key3-$objname $dir/newval || return 1 | |
534 | rm $dir/bad-val $dir/newval | |
535 | ;; | |
536 | ||
537 | 9) | |
538 | objectstore_tool $dir $osd $objname get-attr _ > $dir/robj9-oi | |
539 | echo -n D > $dir/change | |
540 | rados --pool $poolname put $objname $dir/change | |
541 | objectstore_tool $dir $osd $objname set-attr _ $dir/robj9-oi | |
542 | rm $dir/oi $dir/change | |
543 | ;; | |
544 | ||
545 | # ROBJ10 must be handled after digests are re-computed by a deep scrub below | |
546 | # ROBJ11 must be handled with config change before deep scrub | |
547 | # ROBJ12 must be handled with config change before scrubs | |
548 | # ROBJ13 must be handled before scrubs | |
549 | ||
550 | 14) | |
551 | echo -n bad-val > $dir/bad-val | |
552 | objectstore_tool $dir 0 $objname set-attr _ $dir/bad-val || return 1 | |
553 | objectstore_tool $dir 1 $objname rm-attr _ || return 1 | |
554 | rm $dir/bad-val | |
555 | ;; | |
556 | ||
557 | 15) | |
558 | objectstore_tool $dir $osd $objname rm-attr _ || return 1 | |
559 | ||
560 | esac | |
561 | done | |
562 | ||
563 | local pg=$(get_pg $poolname ROBJ0) | |
564 | ||
565 | set_config osd 0 filestore_debug_inject_read_err true || return 1 | |
566 | set_config osd 1 filestore_debug_inject_read_err true || return 1 | |
567 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.1.asok \ | |
568 | injectdataerr $poolname ROBJ11 || return 1 | |
569 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.0.asok \ | |
570 | injectmdataerr $poolname ROBJ12 || return 1 | |
571 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.0.asok \ | |
572 | injectmdataerr $poolname ROBJ13 || return 1 | |
573 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.1.asok \ | |
574 | injectdataerr $poolname ROBJ13 || return 1 | |
575 | ||
576 | pg_scrub $pg | |
577 | ||
578 | rados list-inconsistent-pg $poolname > $dir/json || return 1 | |
579 | # Check pg count | |
580 | test $(jq '. | length' $dir/json) = "1" || return 1 | |
581 | # Check pgid | |
582 | test $(jq -r '.[0]' $dir/json) = $pg || return 1 | |
583 | ||
584 | rados list-inconsistent-obj $pg > $dir/json || return 1 | |
585 | # Get epoch for repair-get requests | |
586 | epoch=$(jq .epoch $dir/json) | |
587 | ||
588 | jq "$jqfilter" << EOF | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/checkcsjson | |
589 | { | |
590 | "inconsistents": [ | |
591 | { | |
592 | "shards": [ | |
593 | { | |
594 | "size": 7, | |
595 | "errors": [], | |
596 | "osd": 0 | |
597 | }, | |
598 | { | |
599 | "size": 9, | |
600 | "errors": [ | |
601 | "size_mismatch_oi" | |
602 | ], | |
603 | "osd": 1 | |
604 | } | |
605 | ], | |
606 | "selected_object_info": "2:ce3f1d6a:::ROBJ1:head(15'3 client.4169.0:1 dirty|omap|data_digest s 7 uv 3 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
607 | "union_shard_errors": [ | |
608 | "size_mismatch_oi" | |
609 | ], | |
610 | "errors": [ | |
611 | "size_mismatch" | |
612 | ], | |
613 | "object": { | |
614 | "version": 3, | |
615 | "snap": "head", | |
616 | "locator": "", | |
617 | "nspace": "", | |
618 | "name": "ROBJ1" | |
619 | } | |
620 | }, | |
621 | { | |
622 | "shards": [ | |
623 | { | |
624 | "errors": [ | |
625 | "stat_error" | |
626 | ], | |
627 | "osd": 0 | |
628 | }, | |
629 | { | |
630 | "size": 7, | |
631 | "errors": [], | |
632 | "osd": 1 | |
633 | } | |
634 | ], | |
635 | "selected_object_info": "2:bc819597:::ROBJ12:head(110'39 client.4732.0:1 dirty|omap|data_digest s 7 uv 39 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
636 | "union_shard_errors": [ | |
637 | "stat_error" | |
638 | ], | |
639 | "errors": [], | |
640 | "object": { | |
641 | "version": 39, | |
642 | "snap": "head", | |
643 | "locator": "", | |
644 | "nspace": "", | |
645 | "name": "ROBJ12" | |
646 | } | |
647 | }, | |
648 | { | |
649 | "shards": [ | |
650 | { | |
651 | "errors": [ | |
652 | "stat_error" | |
653 | ], | |
654 | "osd": 0 | |
655 | }, | |
656 | { | |
657 | "size": 7, | |
658 | "errors": [], | |
659 | "osd": 1 | |
660 | } | |
661 | ], | |
662 | "selected_object_info": "2:d60617f9:::ROBJ13:head(112'42 client.4737.0:1 dirty|omap|data_digest s 7 uv 42 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
663 | "union_shard_errors": [ | |
664 | "stat_error" | |
665 | ], | |
666 | "errors": [], | |
667 | "object": { | |
668 | "version": 42, | |
669 | "snap": "head", | |
670 | "locator": "", | |
671 | "nspace": "", | |
672 | "name": "ROBJ13" | |
673 | } | |
674 | }, | |
675 | { | |
676 | "shards": [ | |
677 | { | |
678 | "size": 7, | |
679 | "errors": [ | |
680 | "oi_attr_corrupted" | |
681 | ], | |
682 | "osd": 0 | |
683 | }, | |
684 | { | |
685 | "size": 7, | |
686 | "errors": [ | |
687 | "oi_attr_missing" | |
688 | ], | |
689 | "osd": 1 | |
690 | } | |
691 | ], | |
692 | "union_shard_errors": [ | |
693 | "oi_attr_missing", | |
694 | "oi_attr_corrupted" | |
695 | ], | |
696 | "errors": [], | |
697 | "object": { | |
698 | "version": 0, | |
699 | "snap": "head", | |
700 | "locator": "", | |
701 | "nspace": "", | |
702 | "name": "ROBJ14" | |
703 | } | |
704 | }, | |
705 | { | |
706 | "shards": [ | |
707 | { | |
708 | "attrs": [ | |
709 | { | |
710 | "Base64": true, | |
711 | "value": "", | |
712 | "name": "_" | |
713 | }, | |
714 | { | |
715 | "Base64": true, | |
716 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
717 | "name": "snapset" | |
718 | } | |
719 | ], | |
720 | "size": 7, | |
721 | "errors": [], | |
722 | "osd": 0 | |
723 | }, | |
724 | { | |
725 | "attrs": [ | |
726 | { | |
727 | "Base64": true, | |
728 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
729 | "name": "snapset" | |
730 | } | |
731 | ], | |
732 | "size": 7, | |
733 | "errors": [ | |
734 | "oi_attr_missing" | |
735 | ], | |
736 | "osd": 1 | |
737 | } | |
738 | ], | |
739 | "selected_object_info": "2:30259878:::ROBJ15:head(127'48 client.4820.0:1 dirty|omap|data_digest s 7 uv 48 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
740 | "union_shard_errors": [ | |
741 | "oi_attr_missing" | |
742 | ], | |
743 | "errors": [ | |
744 | "attr_name_mismatch" | |
745 | ], | |
746 | "object": { | |
747 | "version": 48, | |
748 | "snap": "head", | |
749 | "locator": "", | |
750 | "nspace": "", | |
751 | "name": "ROBJ15" | |
752 | } | |
753 | }, | |
754 | { | |
755 | "shards": [ | |
756 | { | |
757 | "size": 7, | |
758 | "errors": [], | |
759 | "osd": 0 | |
760 | }, | |
761 | { | |
762 | "errors": [ | |
763 | "missing" | |
764 | ], | |
765 | "osd": 1 | |
766 | } | |
767 | ], | |
768 | "selected_object_info": "2:f2a5b2a4:::ROBJ3:head(29'9 client.4251.0:1 dirty|omap|data_digest s 7 uv 9 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
769 | "union_shard_errors": [ | |
770 | "missing" | |
771 | ], | |
772 | "errors": [], | |
773 | "object": { | |
774 | "version": 9, | |
775 | "snap": "head", | |
776 | "locator": "", | |
777 | "nspace": "", | |
778 | "name": "ROBJ3" | |
779 | } | |
780 | }, | |
781 | { | |
782 | "shards": [ | |
783 | { | |
784 | "attrs": [ | |
785 | { | |
786 | "Base64": true, | |
787 | "value": "", | |
788 | "name": "_" | |
789 | }, | |
790 | { | |
791 | "Base64": false, | |
792 | "value": "bad-val", | |
793 | "name": "_key1-ROBJ8" | |
794 | }, | |
795 | { | |
796 | "Base64": false, | |
797 | "value": "val3-ROBJ8", | |
798 | "name": "_key3-ROBJ8" | |
799 | }, | |
800 | { | |
801 | "Base64": true, | |
802 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
803 | "name": "snapset" | |
804 | } | |
805 | ], | |
806 | "size": 7, | |
807 | "errors": [], | |
808 | "osd": 0 | |
809 | }, | |
810 | { | |
811 | "attrs": [ | |
812 | { | |
813 | "Base64": true, | |
814 | "value": "", | |
815 | "name": "_" | |
816 | }, | |
817 | { | |
818 | "Base64": false, | |
819 | "value": "val1-ROBJ8", | |
820 | "name": "_key1-ROBJ8" | |
821 | }, | |
822 | { | |
823 | "Base64": false, | |
824 | "value": "val2-ROBJ8", | |
825 | "name": "_key2-ROBJ8" | |
826 | }, | |
827 | { | |
828 | "Base64": true, | |
829 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
830 | "name": "snapset" | |
831 | } | |
832 | ], | |
833 | "size": 7, | |
834 | "errors": [], | |
835 | "osd": 1 | |
836 | } | |
837 | ], | |
838 | "selected_object_info": "2:86586531:::ROBJ8:head(70'26 client.4495.0:1 dirty|omap|data_digest s 7 uv 26 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
839 | "union_shard_errors": [], | |
840 | "errors": [ | |
841 | "attr_value_mismatch", | |
842 | "attr_name_mismatch" | |
843 | ], | |
844 | "object": { | |
845 | "version": 26, | |
846 | "snap": "head", | |
847 | "locator": "", | |
848 | "nspace": "", | |
849 | "name": "ROBJ8" | |
850 | } | |
851 | }, | |
852 | { | |
853 | "shards": [ | |
854 | { | |
855 | "attrs": [ | |
856 | { | |
857 | "Base64": true, | |
858 | "value": "", | |
859 | "name": "_" | |
860 | }, | |
861 | { | |
862 | "Base64": true, | |
863 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
864 | "name": "snapset" | |
865 | } | |
866 | ], | |
867 | "object_info": "2:ffdb2004:::ROBJ9:head(94'30 client.4649.0:1 dirty|omap|data_digest s 1 uv 30 dd 2b63260d alloc_hint [0 0 0])", | |
868 | "size": 1, | |
869 | "errors": [], | |
870 | "osd": 0 | |
871 | }, | |
872 | { | |
873 | "attrs": [ | |
874 | { | |
875 | "Base64": true, | |
876 | "value": "", | |
877 | "name": "_" | |
878 | }, | |
879 | { | |
880 | "Base64": true, | |
881 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
882 | "name": "snapset" | |
883 | } | |
884 | ], | |
885 | "object_info": "2:ffdb2004:::ROBJ9:head(89'29 client.4612.0:1 dirty|omap|data_digest s 7 uv 29 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
886 | "size": 1, | |
887 | "errors": [], | |
888 | "osd": 1 | |
889 | } | |
890 | ], | |
891 | "selected_object_info": "2:ffdb2004:::ROBJ9:head(94'30 client.4649.0:1 dirty|omap|data_digest s 1 uv 30 dd 2b63260d alloc_hint [0 0 0])", | |
892 | "union_shard_errors": [], | |
893 | "errors": [ | |
894 | "object_info_inconsistency", | |
895 | "attr_value_mismatch" | |
896 | ], | |
897 | "object": { | |
898 | "version": 30, | |
899 | "snap": "head", | |
900 | "locator": "", | |
901 | "nspace": "", | |
902 | "name": "ROBJ9" | |
903 | } | |
904 | } | |
905 | ], | |
906 | "epoch": 0 | |
907 | } | |
908 | EOF | |
909 | ||
910 | jq "$jqfilter" $dir/json | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/csjson | |
911 | diff -y $termwidth $dir/checkcsjson $dir/csjson || test $getjson = "yes" || return 1 | |
912 | if test $getjson = "yes" | |
913 | then | |
914 | jq '.' $dir/json > save1.json | |
915 | fi | |
916 | ||
917 | if which jsonschema > /dev/null; | |
918 | then | |
919 | jsonschema -i $dir/json $CEPH_ROOT/doc/rados/command/list-inconsistent-obj.json || return 1 | |
920 | fi | |
921 | ||
922 | # Compute an old omap digest and save oi | |
923 | CEPH_ARGS='' ceph daemon $dir//ceph-osd.0.asok \ | |
924 | config set osd_deep_scrub_update_digest_min_age 0 | |
925 | CEPH_ARGS='' ceph daemon $dir//ceph-osd.1.asok \ | |
926 | config set osd_deep_scrub_update_digest_min_age 0 | |
927 | pg_deep_scrub $pg | |
928 | ||
929 | objname=ROBJ9 | |
930 | # Change data and size again because digest was recomputed | |
931 | echo -n ZZZ > $dir/change | |
932 | rados --pool $poolname put $objname $dir/change | |
933 | # Set one to an even older value | |
934 | objectstore_tool $dir 0 $objname set-attr _ $dir/robj9-oi | |
935 | rm $dir/oi $dir/change | |
936 | ||
937 | objname=ROBJ10 | |
938 | objectstore_tool $dir 1 $objname get-attr _ > $dir/oi | |
939 | rados --pool $poolname setomapval $objname key2-$objname val2-$objname | |
940 | objectstore_tool $dir 0 $objname set-attr _ $dir/oi | |
941 | objectstore_tool $dir 1 $objname set-attr _ $dir/oi | |
942 | rm $dir/oi | |
943 | ||
944 | set_config osd 0 filestore_debug_inject_read_err true || return 1 | |
945 | set_config osd 1 filestore_debug_inject_read_err true || return 1 | |
946 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.1.asok \ | |
947 | injectdataerr $poolname ROBJ11 || return 1 | |
948 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.0.asok \ | |
949 | injectmdataerr $poolname ROBJ12 || return 1 | |
950 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.0.asok \ | |
951 | injectmdataerr $poolname ROBJ13 || return 1 | |
952 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.1.asok \ | |
953 | injectdataerr $poolname ROBJ13 || return 1 | |
954 | pg_deep_scrub $pg | |
955 | ||
956 | rados list-inconsistent-pg $poolname > $dir/json || return 1 | |
957 | # Check pg count | |
958 | test $(jq '. | length' $dir/json) = "1" || return 1 | |
959 | # Check pgid | |
960 | test $(jq -r '.[0]' $dir/json) = $pg || return 1 | |
961 | ||
962 | rados list-inconsistent-obj $pg > $dir/json || return 1 | |
963 | # Get epoch for repair-get requests | |
964 | epoch=$(jq .epoch $dir/json) | |
965 | ||
966 | jq "$jqfilter" << EOF | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/checkcsjson | |
967 | { | |
968 | "inconsistents": [ | |
969 | { | |
970 | "shards": [ | |
971 | { | |
972 | "data_digest": "0x2ddbf8f5", | |
973 | "omap_digest": "0xf5fba2c6", | |
974 | "size": 7, | |
975 | "errors": [], | |
976 | "osd": 0 | |
977 | }, | |
978 | { | |
979 | "data_digest": "0x2d4a11c2", | |
980 | "omap_digest": "0xf5fba2c6", | |
981 | "size": 9, | |
982 | "errors": [ | |
983 | "data_digest_mismatch_oi", | |
984 | "size_mismatch_oi" | |
985 | ], | |
986 | "osd": 1 | |
987 | } | |
988 | ], | |
989 | "selected_object_info": "2:ce3f1d6a:::ROBJ1:head(15'3 client.4171.0:1 dirty|omap|data_digest s 7 uv 3 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
990 | "union_shard_errors": [ | |
991 | "data_digest_mismatch_oi", | |
992 | "size_mismatch_oi" | |
993 | ], | |
994 | "errors": [ | |
995 | "data_digest_mismatch", | |
996 | "size_mismatch" | |
997 | ], | |
998 | "object": { | |
999 | "version": 3, | |
1000 | "snap": "head", | |
1001 | "locator": "", | |
1002 | "nspace": "", | |
1003 | "name": "ROBJ1" | |
1004 | } | |
1005 | }, | |
1006 | { | |
1007 | "shards": [ | |
1008 | { | |
1009 | "data_digest": "0x2ddbf8f5", | |
1010 | "omap_digest": "0xa8dd5adc", | |
1011 | "size": 7, | |
1012 | "errors": [ | |
1013 | "omap_digest_mismatch_oi" | |
1014 | ], | |
1015 | "osd": 0 | |
1016 | }, | |
1017 | { | |
1018 | "data_digest": "0x2ddbf8f5", | |
1019 | "omap_digest": "0xa8dd5adc", | |
1020 | "size": 7, | |
1021 | "errors": [ | |
1022 | "omap_digest_mismatch_oi" | |
1023 | ], | |
1024 | "osd": 1 | |
1025 | } | |
1026 | ], | |
1027 | "selected_object_info": "2:b1f19cbd:::ROBJ10:head(136'51 osd.0.0:8 dirty|omap|data_digest|omap_digest s 7 uv 33 dd 2ddbf8f5 od c2025a24 alloc_hint [0 0 0])", | |
1028 | "union_shard_errors": [ | |
1029 | "omap_digest_mismatch_oi" | |
1030 | ], | |
1031 | "errors": [], | |
1032 | "object": { | |
1033 | "version": 33, | |
1034 | "snap": "head", | |
1035 | "locator": "", | |
1036 | "nspace": "", | |
1037 | "name": "ROBJ10" | |
1038 | } | |
1039 | }, | |
1040 | { | |
1041 | "shards": [ | |
1042 | { | |
1043 | "data_digest": "0x2ddbf8f5", | |
1044 | "omap_digest": "0xa03cef03", | |
1045 | "size": 7, | |
1046 | "errors": [], | |
1047 | "osd": 0 | |
1048 | }, | |
1049 | { | |
1050 | "size": 7, | |
1051 | "errors": [ | |
1052 | "read_error" | |
1053 | ], | |
1054 | "osd": 1 | |
1055 | } | |
1056 | ], | |
1057 | "selected_object_info": "2:87abbf36:::ROBJ11:head(105'36 client.4699.0:1 dirty|omap|data_digest s 7 uv 36 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1058 | "union_shard_errors": [ | |
1059 | "read_error" | |
1060 | ], | |
1061 | "errors": [], | |
1062 | "object": { | |
1063 | "version": 36, | |
1064 | "snap": "head", | |
1065 | "locator": "", | |
1066 | "nspace": "", | |
1067 | "name": "ROBJ11" | |
1068 | } | |
1069 | }, | |
1070 | { | |
1071 | "shards": [ | |
1072 | { | |
1073 | "errors": [ | |
1074 | "stat_error" | |
1075 | ], | |
1076 | "osd": 0 | |
1077 | }, | |
1078 | { | |
1079 | "data_digest": "0x2ddbf8f5", | |
1080 | "omap_digest": "0x067f306a", | |
1081 | "size": 7, | |
1082 | "errors": [], | |
1083 | "osd": 1 | |
1084 | } | |
1085 | ], | |
1086 | "selected_object_info": "2:bc819597:::ROBJ12:head(107'39 client.4704.0:1 dirty|omap|data_digest s 7 uv 39 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1087 | "union_shard_errors": [ | |
1088 | "stat_error" | |
1089 | ], | |
1090 | "errors": [], | |
1091 | "object": { | |
1092 | "version": 39, | |
1093 | "snap": "head", | |
1094 | "locator": "", | |
1095 | "nspace": "", | |
1096 | "name": "ROBJ12" | |
1097 | } | |
1098 | }, | |
1099 | { | |
1100 | "shards": [ | |
1101 | { | |
1102 | "errors": [ | |
1103 | "stat_error" | |
1104 | ], | |
1105 | "osd": 0 | |
1106 | }, | |
1107 | { | |
1108 | "size": 7, | |
1109 | "errors": [ | |
1110 | "read_error" | |
1111 | ], | |
1112 | "osd": 1 | |
1113 | } | |
1114 | ], | |
1115 | "union_shard_errors": [ | |
1116 | "stat_error", | |
1117 | "read_error" | |
1118 | ], | |
1119 | "errors": [], | |
1120 | "object": { | |
1121 | "version": 0, | |
1122 | "snap": "head", | |
1123 | "locator": "", | |
1124 | "nspace": "", | |
1125 | "name": "ROBJ13" | |
1126 | } | |
1127 | }, | |
1128 | { | |
1129 | "shards": [ | |
1130 | { | |
1131 | "data_digest": "0x2ddbf8f5", | |
1132 | "omap_digest": "0x4f14f849", | |
1133 | "size": 7, | |
1134 | "errors": [ | |
1135 | "oi_attr_corrupted" | |
1136 | ], | |
1137 | "osd": 0 | |
1138 | }, | |
1139 | { | |
1140 | "data_digest": "0x2ddbf8f5", | |
1141 | "omap_digest": "0x4f14f849", | |
1142 | "size": 7, | |
1143 | "errors": [ | |
1144 | "oi_attr_missing" | |
1145 | ], | |
1146 | "osd": 1 | |
1147 | } | |
1148 | ], | |
1149 | "union_shard_errors": [ | |
1150 | "oi_attr_missing", | |
1151 | "oi_attr_corrupted" | |
1152 | ], | |
1153 | "errors": [], | |
1154 | "object": { | |
1155 | "version": 0, | |
1156 | "snap": "head", | |
1157 | "locator": "", | |
1158 | "nspace": "", | |
1159 | "name": "ROBJ14" | |
1160 | } | |
1161 | }, | |
1162 | { | |
1163 | "shards": [ | |
1164 | { | |
1165 | "attrs": [ | |
1166 | { | |
1167 | "Base64": true, | |
1168 | "value": "", | |
1169 | "name": "_" | |
1170 | }, | |
1171 | { | |
1172 | "Base64": true, | |
1173 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1174 | "name": "snapset" | |
1175 | } | |
1176 | ], | |
1177 | "data_digest": "0x2ddbf8f5", | |
1178 | "omap_digest": "0x2d2a4d6e", | |
1179 | "size": 7, | |
1180 | "errors": [], | |
1181 | "osd": 0 | |
1182 | }, | |
1183 | { | |
1184 | "attrs": [ | |
1185 | { | |
1186 | "Base64": true, | |
1187 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1188 | "name": "snapset" | |
1189 | } | |
1190 | ], | |
1191 | "data_digest": "0x2ddbf8f5", | |
1192 | "omap_digest": "0x2d2a4d6e", | |
1193 | "size": 7, | |
1194 | "errors": [ | |
1195 | "oi_attr_missing" | |
1196 | ], | |
1197 | "osd": 1 | |
1198 | } | |
1199 | ], | |
1200 | "selected_object_info": "2:30259878:::ROBJ15:head(124'48 client.4792.0:1 dirty|omap|data_digest s 7 uv 48 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1201 | "union_shard_errors": [ | |
1202 | "oi_attr_missing" | |
1203 | ], | |
1204 | "errors": [ | |
1205 | "attr_name_mismatch" | |
1206 | ], | |
1207 | "object": { | |
1208 | "version": 48, | |
1209 | "snap": "head", | |
1210 | "locator": "", | |
1211 | "nspace": "", | |
1212 | "name": "ROBJ15" | |
1213 | } | |
1214 | }, | |
1215 | { | |
1216 | "shards": [ | |
1217 | { | |
1218 | "data_digest": "0x578a4830", | |
1219 | "omap_digest": "0xf8e11918", | |
1220 | "size": 7, | |
1221 | "errors": [ | |
1222 | "data_digest_mismatch_oi" | |
1223 | ], | |
1224 | "osd": 0 | |
1225 | }, | |
1226 | { | |
1227 | "data_digest": "0x2ddbf8f5", | |
1228 | "omap_digest": "0xf8e11918", | |
1229 | "size": 7, | |
1230 | "errors": [], | |
1231 | "osd": 1 | |
1232 | } | |
1233 | ], | |
1234 | "selected_object_info": "2:e97ce31e:::ROBJ2:head(22'6 client.4214.0:1 dirty|omap|data_digest s 7 uv 6 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1235 | "union_shard_errors": [ | |
1236 | "data_digest_mismatch_oi" | |
1237 | ], | |
1238 | "errors": [ | |
1239 | "data_digest_mismatch" | |
1240 | ], | |
1241 | "object": { | |
1242 | "version": 6, | |
1243 | "snap": "head", | |
1244 | "locator": "", | |
1245 | "nspace": "", | |
1246 | "name": "ROBJ2" | |
1247 | } | |
1248 | }, | |
1249 | { | |
1250 | "shards": [ | |
1251 | { | |
1252 | "data_digest": "0x2ddbf8f5", | |
1253 | "omap_digest": "0x00b35dfd", | |
1254 | "size": 7, | |
1255 | "errors": [], | |
1256 | "osd": 0 | |
1257 | }, | |
1258 | { | |
1259 | "errors": [ | |
1260 | "missing" | |
1261 | ], | |
1262 | "osd": 1 | |
1263 | } | |
1264 | ], | |
1265 | "selected_object_info": "2:f2a5b2a4:::ROBJ3:head(29'9 client.4255.0:1 dirty|omap|data_digest s 7 uv 9 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1266 | "union_shard_errors": [ | |
1267 | "missing" | |
1268 | ], | |
1269 | "errors": [], | |
1270 | "object": { | |
1271 | "version": 9, | |
1272 | "snap": "head", | |
1273 | "locator": "", | |
1274 | "nspace": "", | |
1275 | "name": "ROBJ3" | |
1276 | } | |
1277 | }, | |
1278 | { | |
1279 | "shards": [ | |
1280 | { | |
1281 | "data_digest": "0x2ddbf8f5", | |
1282 | "omap_digest": "0xd7178dfe", | |
1283 | "size": 7, | |
1284 | "errors": [], | |
1285 | "osd": 0 | |
1286 | }, | |
1287 | { | |
1288 | "data_digest": "0x2ddbf8f5", | |
1289 | "omap_digest": "0xe2d46ea4", | |
1290 | "size": 7, | |
1291 | "errors": [ | |
1292 | "omap_digest_mismatch_oi" | |
1293 | ], | |
1294 | "osd": 1 | |
1295 | } | |
1296 | ], | |
1297 | "selected_object_info": "2:f4981d31:::ROBJ4:head(136'52 osd.0.0:9 dirty|omap|data_digest|omap_digest s 7 uv 12 dd 2ddbf8f5 od d7178dfe alloc_hint [0 0 0])", | |
1298 | "union_shard_errors": [ | |
1299 | "omap_digest_mismatch_oi" | |
1300 | ], | |
1301 | "errors": [ | |
1302 | "omap_digest_mismatch" | |
1303 | ], | |
1304 | "object": { | |
1305 | "version": 12, | |
1306 | "snap": "head", | |
1307 | "locator": "", | |
1308 | "nspace": "", | |
1309 | "name": "ROBJ4" | |
1310 | } | |
1311 | }, | |
1312 | { | |
1313 | "shards": [ | |
1314 | { | |
1315 | "data_digest": "0x2ddbf8f5", | |
1316 | "omap_digest": "0x1a862a41", | |
1317 | "size": 7, | |
1318 | "errors": [], | |
1319 | "osd": 0 | |
1320 | }, | |
1321 | { | |
1322 | "data_digest": "0x2ddbf8f5", | |
1323 | "omap_digest": "0x06cac8f6", | |
1324 | "size": 7, | |
1325 | "errors": [ | |
1326 | "omap_digest_mismatch_oi" | |
1327 | ], | |
1328 | "osd": 1 | |
1329 | } | |
1330 | ], | |
1331 | "selected_object_info": "2:f4bfd4d1:::ROBJ5:head(136'53 osd.0.0:10 dirty|omap|data_digest|omap_digest s 7 uv 15 dd 2ddbf8f5 od 1a862a41 alloc_hint [0 0 0])", | |
1332 | "union_shard_errors": [ | |
1333 | "omap_digest_mismatch_oi" | |
1334 | ], | |
1335 | "errors": [ | |
1336 | "omap_digest_mismatch" | |
1337 | ], | |
1338 | "object": { | |
1339 | "version": 15, | |
1340 | "snap": "head", | |
1341 | "locator": "", | |
1342 | "nspace": "", | |
1343 | "name": "ROBJ5" | |
1344 | } | |
1345 | }, | |
1346 | { | |
1347 | "shards": [ | |
1348 | { | |
1349 | "data_digest": "0x2ddbf8f5", | |
1350 | "omap_digest": "0x689ee887", | |
1351 | "size": 7, | |
1352 | "errors": [], | |
1353 | "osd": 0 | |
1354 | }, | |
1355 | { | |
1356 | "data_digest": "0x2ddbf8f5", | |
1357 | "omap_digest": "0x179c919f", | |
1358 | "size": 7, | |
1359 | "errors": [ | |
1360 | "omap_digest_mismatch_oi" | |
1361 | ], | |
1362 | "osd": 1 | |
1363 | } | |
1364 | ], | |
1365 | "selected_object_info": "2:a53c12e8:::ROBJ6:head(136'50 osd.0.0:7 dirty|omap|data_digest|omap_digest s 7 uv 18 dd 2ddbf8f5 od 689ee887 alloc_hint [0 0 0])", | |
1366 | "union_shard_errors": [ | |
1367 | "omap_digest_mismatch_oi" | |
1368 | ], | |
1369 | "errors": [ | |
1370 | "omap_digest_mismatch" | |
1371 | ], | |
1372 | "object": { | |
1373 | "version": 18, | |
1374 | "snap": "head", | |
1375 | "locator": "", | |
1376 | "nspace": "", | |
1377 | "name": "ROBJ6" | |
1378 | } | |
1379 | }, | |
1380 | { | |
1381 | "shards": [ | |
1382 | { | |
1383 | "data_digest": "0x2ddbf8f5", | |
1384 | "omap_digest": "0xefced57a", | |
1385 | "size": 7, | |
1386 | "errors": [], | |
1387 | "osd": 0 | |
1388 | }, | |
1389 | { | |
1390 | "data_digest": "0x2ddbf8f5", | |
1391 | "omap_digest": "0x6a73cc07", | |
1392 | "size": 7, | |
1393 | "errors": [ | |
1394 | "omap_digest_mismatch_oi" | |
1395 | ], | |
1396 | "osd": 1 | |
1397 | } | |
1398 | ], | |
1399 | "selected_object_info": "2:8b55fa4b:::ROBJ7:head(123'50 osd.0.0:7 dirty|omap|data_digest|omap_digest s 7 uv 21 dd 2ddbf8f5 od efced57a alloc_hint [0 0 0])", | |
1400 | "union_shard_errors": [ | |
1401 | "omap_digest_mismatch_oi" | |
1402 | ], | |
1403 | "errors": [ | |
1404 | "omap_digest_mismatch" | |
1405 | ], | |
1406 | "object": { | |
1407 | "version": 21, | |
1408 | "snap": "head", | |
1409 | "locator": "", | |
1410 | "nspace": "", | |
1411 | "name": "ROBJ7" | |
1412 | } | |
1413 | }, | |
1414 | { | |
1415 | "shards": [ | |
1416 | { | |
1417 | "attrs": [ | |
1418 | { | |
1419 | "Base64": true, | |
1420 | "value": "", | |
1421 | "name": "_" | |
1422 | }, | |
1423 | { | |
1424 | "Base64": false, | |
1425 | "value": "bad-val", | |
1426 | "name": "_key1-ROBJ8" | |
1427 | }, | |
1428 | { | |
1429 | "Base64": false, | |
1430 | "value": "val3-ROBJ8", | |
1431 | "name": "_key3-ROBJ8" | |
1432 | }, | |
1433 | { | |
1434 | "Base64": true, | |
1435 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1436 | "name": "snapset" | |
1437 | } | |
1438 | ], | |
1439 | "data_digest": "0x2ddbf8f5", | |
1440 | "omap_digest": "0xd6be81dc", | |
1441 | "size": 7, | |
1442 | "errors": [], | |
1443 | "osd": 0 | |
1444 | }, | |
1445 | { | |
1446 | "attrs": [ | |
1447 | { | |
1448 | "Base64": true, | |
1449 | "value": "", | |
1450 | "name": "_" | |
1451 | }, | |
1452 | { | |
1453 | "Base64": false, | |
1454 | "value": "val1-ROBJ8", | |
1455 | "name": "_key1-ROBJ8" | |
1456 | }, | |
1457 | { | |
1458 | "Base64": false, | |
1459 | "value": "val2-ROBJ8", | |
1460 | "name": "_key2-ROBJ8" | |
1461 | }, | |
1462 | { | |
1463 | "Base64": true, | |
1464 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1465 | "name": "snapset" | |
1466 | } | |
1467 | ], | |
1468 | "data_digest": "0x2ddbf8f5", | |
1469 | "omap_digest": "0xd6be81dc", | |
1470 | "size": 7, | |
1471 | "errors": [], | |
1472 | "osd": 1 | |
1473 | } | |
1474 | ], | |
1475 | "selected_object_info": "2:86586531:::ROBJ8:head(136'49 osd.0.0:6 dirty|omap|data_digest|omap_digest s 7 uv 26 dd 2ddbf8f5 od d6be81dc alloc_hint [0 0 0])", | |
1476 | "union_shard_errors": [], | |
1477 | "errors": [ | |
1478 | "attr_value_mismatch", | |
1479 | "attr_name_mismatch" | |
1480 | ], | |
1481 | "object": { | |
1482 | "version": 26, | |
1483 | "snap": "head", | |
1484 | "locator": "", | |
1485 | "nspace": "", | |
1486 | "name": "ROBJ8" | |
1487 | } | |
1488 | }, | |
1489 | { | |
1490 | "shards": [ | |
1491 | { | |
1492 | "attrs": [ | |
1493 | { | |
1494 | "Base64": true, | |
1495 | "value": "", | |
1496 | "name": "_" | |
1497 | }, | |
1498 | { | |
1499 | "Base64": true, | |
1500 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1501 | "name": "snapset" | |
1502 | } | |
1503 | ], | |
1504 | "object_info": "2:ffdb2004:::ROBJ9:head(90'29 client.4615.0:1 dirty|omap|data_digest s 7 uv 29 dd 2ddbf8f5 alloc_hint [0 0 0])", | |
1505 | "data_digest": "0x1f26fb26", | |
1506 | "omap_digest": "0x2eecc539", | |
1507 | "size": 3, | |
1508 | "errors": [], | |
1509 | "osd": 0 | |
1510 | }, | |
1511 | { | |
1512 | "attrs": [ | |
1513 | { | |
1514 | "Base64": true, | |
1515 | "value": "", | |
1516 | "name": "_" | |
1517 | }, | |
1518 | { | |
1519 | "Base64": true, | |
1520 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1521 | "name": "snapset" | |
1522 | } | |
1523 | ], | |
1524 | "object_info": "2:ffdb2004:::ROBJ9:head(123'56 client.4891.0:1 dirty|omap|data_digest|omap_digest s 3 uv 56 dd 1f26fb26 od 2eecc539 alloc_hint [0 0 0])", | |
1525 | "data_digest": "0x1f26fb26", | |
1526 | "omap_digest": "0x2eecc539", | |
1527 | "size": 3, | |
1528 | "errors": [], | |
1529 | "osd": 1 | |
1530 | } | |
1531 | ], | |
1532 | "selected_object_info": "2:ffdb2004:::ROBJ9:head(123'56 client.4891.0:1 dirty|omap|data_digest|omap_digest s 3 uv 56 dd 1f26fb26 od 2eecc539 alloc_hint [0 0 0])", | |
1533 | "union_shard_errors": [], | |
1534 | "errors": [ | |
1535 | "object_info_inconsistency", | |
1536 | "attr_value_mismatch" | |
1537 | ], | |
1538 | "object": { | |
1539 | "version": 56, | |
1540 | "snap": "head", | |
1541 | "locator": "", | |
1542 | "nspace": "", | |
1543 | "name": "ROBJ9" | |
1544 | } | |
1545 | } | |
1546 | ], | |
1547 | "epoch": 0 | |
1548 | } | |
1549 | EOF | |
1550 | ||
1551 | jq "$jqfilter" $dir/json | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/csjson | |
1552 | diff -y $termwidth $dir/checkcsjson $dir/csjson || test $getjson = "yes" || return 1 | |
1553 | if test $getjson = "yes" | |
1554 | then | |
1555 | jq '.' $dir/json > save2.json | |
1556 | fi | |
1557 | ||
1558 | if which jsonschema > /dev/null; | |
1559 | then | |
1560 | jsonschema -i $dir/json $CEPH_ROOT/doc/rados/command/list-inconsistent-obj.json || return 1 | |
1561 | fi | |
1562 | ||
1563 | rados rmpool $poolname $poolname --yes-i-really-really-mean-it | |
1564 | teardown $dir || return 1 | |
1565 | } | |
1566 | ||
1567 | ||
1568 | # | |
1569 | # Test scrub errors for an erasure coded pool | |
1570 | # | |
1571 | function corrupt_scrub_erasure() { | |
1572 | local dir=$1 | |
1573 | local allow_overwrites=$2 | |
1574 | local poolname=ecpool | |
1575 | local total_objs=5 | |
1576 | ||
1577 | setup $dir || return 1 | |
1578 | run_mon $dir a || return 1 | |
1579 | run_mgr $dir x || return 1 | |
1580 | for id in $(seq 0 2) ; do | |
1581 | if [ "$allow_overwrites" = "true" ]; then | |
1582 | run_osd_bluestore $dir $id || return 1 | |
1583 | else | |
1584 | run_osd $dir $id || return 1 | |
1585 | fi | |
1586 | done | |
1587 | wait_for_clean || return 1 | |
1588 | ||
1589 | create_ec_pool $poolname $allow_overwrites k=2 m=1 stripe_unit=2K --force || return 1 | |
1590 | ||
1591 | for i in $(seq 1 $total_objs) ; do | |
1592 | objname=EOBJ${i} | |
1593 | add_something $dir $poolname $objname || return 1 | |
1594 | ||
1595 | local osd=$(expr $i % 2) | |
1596 | ||
1597 | case $i in | |
1598 | 1) | |
1599 | # Size (deep scrub data_digest too) | |
1600 | local payload=UVWXYZZZ | |
1601 | echo $payload > $dir/CORRUPT | |
1602 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
1603 | ;; | |
1604 | ||
1605 | 2) | |
1606 | # Corrupt EC shard | |
1607 | dd if=/dev/urandom of=$dir/CORRUPT bs=2048 count=1 | |
1608 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
1609 | ;; | |
1610 | ||
1611 | 3) | |
1612 | # missing | |
1613 | objectstore_tool $dir $osd $objname remove || return 1 | |
1614 | ;; | |
1615 | ||
1616 | 4) | |
1617 | rados --pool $poolname setxattr $objname key1-$objname val1-$objname || return 1 | |
1618 | rados --pool $poolname setxattr $objname key2-$objname val2-$objname || return 1 | |
1619 | ||
1620 | # Break xattrs | |
1621 | echo -n bad-val > $dir/bad-val | |
1622 | objectstore_tool $dir $osd $objname set-attr _key1-$objname $dir/bad-val || return 1 | |
1623 | objectstore_tool $dir $osd $objname rm-attr _key2-$objname || return 1 | |
1624 | echo -n val3-$objname > $dir/newval | |
1625 | objectstore_tool $dir $osd $objname set-attr _key3-$objname $dir/newval || return 1 | |
1626 | rm $dir/bad-val $dir/newval | |
1627 | ;; | |
1628 | ||
1629 | 5) | |
1630 | # Corrupt EC shard | |
1631 | dd if=/dev/urandom of=$dir/CORRUPT bs=2048 count=2 | |
1632 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
1633 | ;; | |
1634 | ||
1635 | esac | |
1636 | done | |
1637 | ||
1638 | local pg=$(get_pg $poolname EOBJ0) | |
1639 | ||
1640 | pg_scrub $pg | |
1641 | ||
1642 | rados list-inconsistent-pg $poolname > $dir/json || return 1 | |
1643 | # Check pg count | |
1644 | test $(jq '. | length' $dir/json) = "1" || return 1 | |
1645 | # Check pgid | |
1646 | test $(jq -r '.[0]' $dir/json) = $pg || return 1 | |
1647 | ||
1648 | rados list-inconsistent-obj $pg > $dir/json || return 1 | |
1649 | # Get epoch for repair-get requests | |
1650 | epoch=$(jq .epoch $dir/json) | |
1651 | ||
1652 | jq "$jqfilter" << EOF | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/checkcsjson | |
1653 | { | |
1654 | "inconsistents": [ | |
1655 | { | |
1656 | "shards": [ | |
1657 | { | |
1658 | "size": 2048, | |
1659 | "errors": [], | |
1660 | "shard": 2, | |
1661 | "osd": 0 | |
1662 | }, | |
1663 | { | |
1664 | "size": 9, | |
1665 | "errors": [ | |
1666 | "size_mismatch_oi" | |
1667 | ], | |
1668 | "shard": 1, | |
1669 | "osd": 1 | |
1670 | }, | |
1671 | { | |
1672 | "size": 2048, | |
1673 | "errors": [], | |
1674 | "shard": 0, | |
1675 | "osd": 2 | |
1676 | } | |
1677 | ], | |
1678 | "selected_object_info": "2:9175b684:::EOBJ1:head(21'1 client.4179.0:1 dirty|data_digest|omap_digest s 7 uv 1 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1679 | "union_shard_errors": [ | |
1680 | "size_mismatch_oi" | |
1681 | ], | |
1682 | "errors": [ | |
1683 | "size_mismatch" | |
1684 | ], | |
1685 | "object": { | |
1686 | "version": 1, | |
1687 | "snap": "head", | |
1688 | "locator": "", | |
1689 | "nspace": "", | |
1690 | "name": "EOBJ1" | |
1691 | } | |
1692 | }, | |
1693 | { | |
1694 | "shards": [ | |
1695 | { | |
1696 | "size": 2048, | |
1697 | "errors": [], | |
1698 | "shard": 2, | |
1699 | "osd": 0 | |
1700 | }, | |
1701 | { | |
1702 | "errors": [ | |
1703 | "missing" | |
1704 | ], | |
1705 | "shard": 1, | |
1706 | "osd": 1 | |
1707 | }, | |
1708 | { | |
1709 | "size": 2048, | |
1710 | "errors": [], | |
1711 | "shard": 0, | |
1712 | "osd": 2 | |
1713 | } | |
1714 | ], | |
1715 | "selected_object_info": "2:b197b25d:::EOBJ3:head(37'3 client.4251.0:1 dirty|data_digest|omap_digest s 7 uv 3 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1716 | "union_shard_errors": [ | |
1717 | "missing" | |
1718 | ], | |
1719 | "errors": [], | |
1720 | "object": { | |
1721 | "version": 3, | |
1722 | "snap": "head", | |
1723 | "locator": "", | |
1724 | "nspace": "", | |
1725 | "name": "EOBJ3" | |
1726 | } | |
1727 | }, | |
1728 | { | |
1729 | "shards": [ | |
1730 | { | |
1731 | "attrs": [ | |
1732 | { | |
1733 | "Base64": true, | |
1734 | "value": "", | |
1735 | "name": "_" | |
1736 | }, | |
1737 | { | |
1738 | "Base64": false, | |
1739 | "value": "bad-val", | |
1740 | "name": "_key1-EOBJ4" | |
1741 | }, | |
1742 | { | |
1743 | "Base64": false, | |
1744 | "value": "val3-EOBJ4", | |
1745 | "name": "_key3-EOBJ4" | |
1746 | }, | |
1747 | { | |
1748 | "Base64": true, | |
1749 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
1750 | "name": "hinfo_key" | |
1751 | }, | |
1752 | { | |
1753 | "Base64": true, | |
1754 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1755 | "name": "snapset" | |
1756 | } | |
1757 | ], | |
1758 | "size": 2048, | |
1759 | "errors": [], | |
1760 | "shard": 2, | |
1761 | "osd": 0 | |
1762 | }, | |
1763 | { | |
1764 | "attrs": [ | |
1765 | { | |
1766 | "Base64": true, | |
1767 | "value": "", | |
1768 | "name": "_" | |
1769 | }, | |
1770 | { | |
1771 | "Base64": false, | |
1772 | "value": "val1-EOBJ4", | |
1773 | "name": "_key1-EOBJ4" | |
1774 | }, | |
1775 | { | |
1776 | "Base64": false, | |
1777 | "value": "val2-EOBJ4", | |
1778 | "name": "_key2-EOBJ4" | |
1779 | }, | |
1780 | { | |
1781 | "Base64": true, | |
1782 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
1783 | "name": "hinfo_key" | |
1784 | }, | |
1785 | { | |
1786 | "Base64": true, | |
1787 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1788 | "name": "snapset" | |
1789 | } | |
1790 | ], | |
1791 | "size": 2048, | |
1792 | "errors": [], | |
1793 | "shard": 1, | |
1794 | "osd": 1 | |
1795 | }, | |
1796 | { | |
1797 | "attrs": [ | |
1798 | { | |
1799 | "Base64": true, | |
1800 | "value": "", | |
1801 | "name": "_" | |
1802 | }, | |
1803 | { | |
1804 | "Base64": false, | |
1805 | "value": "val1-EOBJ4", | |
1806 | "name": "_key1-EOBJ4" | |
1807 | }, | |
1808 | { | |
1809 | "Base64": false, | |
1810 | "value": "val2-EOBJ4", | |
1811 | "name": "_key2-EOBJ4" | |
1812 | }, | |
1813 | { | |
1814 | "Base64": true, | |
1815 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
1816 | "name": "hinfo_key" | |
1817 | }, | |
1818 | { | |
1819 | "Base64": true, | |
1820 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
1821 | "name": "snapset" | |
1822 | } | |
1823 | ], | |
1824 | "size": 2048, | |
1825 | "errors": [], | |
1826 | "shard": 0, | |
1827 | "osd": 2 | |
1828 | } | |
1829 | ], | |
1830 | "selected_object_info": "2:5e723e06:::EOBJ4:head(45'6 client.4289.0:1 dirty|data_digest|omap_digest s 7 uv 6 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1831 | "union_shard_errors": [], | |
1832 | "errors": [ | |
1833 | "attr_value_mismatch", | |
1834 | "attr_name_mismatch" | |
1835 | ], | |
1836 | "object": { | |
1837 | "version": 6, | |
1838 | "snap": "head", | |
1839 | "locator": "", | |
1840 | "nspace": "", | |
1841 | "name": "EOBJ4" | |
1842 | } | |
1843 | }, | |
1844 | { | |
1845 | "shards": [ | |
1846 | { | |
1847 | "size": 2048, | |
1848 | "errors": [], | |
1849 | "shard": 2, | |
1850 | "osd": 0 | |
1851 | }, | |
1852 | { | |
1853 | "size": 4096, | |
1854 | "errors": [ | |
1855 | "size_mismatch_oi" | |
1856 | ], | |
1857 | "shard": 1, | |
1858 | "osd": 1 | |
1859 | }, | |
1860 | { | |
1861 | "size": 2048, | |
1862 | "errors": [], | |
1863 | "shard": 0, | |
1864 | "osd": 2 | |
1865 | } | |
1866 | ], | |
1867 | "selected_object_info": "2:8549dfb5:::EOBJ5:head(65'7 client.4441.0:1 dirty|data_digest|omap_digest s 7 uv 7 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1868 | "union_shard_errors": [ | |
1869 | "size_mismatch_oi" | |
1870 | ], | |
1871 | "errors": [ | |
1872 | "size_mismatch" | |
1873 | ], | |
1874 | "object": { | |
1875 | "version": 7, | |
1876 | "snap": "head", | |
1877 | "locator": "", | |
1878 | "nspace": "", | |
1879 | "name": "EOBJ5" | |
1880 | } | |
1881 | } | |
1882 | ], | |
1883 | "epoch": 0 | |
1884 | } | |
1885 | EOF | |
1886 | ||
1887 | jq "$jqfilter" $dir/json | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/csjson | |
1888 | diff -y $termwidth $dir/checkcsjson $dir/csjson || test $getjson = "yes" || return 1 | |
1889 | if test $getjson = "yes" | |
1890 | then | |
1891 | jq '.' $dir/json > save3.json | |
1892 | fi | |
1893 | ||
1894 | if which jsonschema > /dev/null; | |
1895 | then | |
1896 | jsonschema -i $dir/json $CEPH_ROOT/doc/rados/command/list-inconsistent-obj.json || return 1 | |
1897 | fi | |
1898 | ||
1899 | pg_deep_scrub $pg | |
1900 | ||
1901 | rados list-inconsistent-pg $poolname > $dir/json || return 1 | |
1902 | # Check pg count | |
1903 | test $(jq '. | length' $dir/json) = "1" || return 1 | |
1904 | # Check pgid | |
1905 | test $(jq -r '.[0]' $dir/json) = $pg || return 1 | |
1906 | ||
1907 | rados list-inconsistent-obj $pg > $dir/json || return 1 | |
1908 | # Get epoch for repair-get requests | |
1909 | epoch=$(jq .epoch $dir/json) | |
1910 | ||
1911 | jq "$jqfilter" << EOF | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/checkcsjson | |
1912 | { | |
1913 | "inconsistents": [ | |
1914 | { | |
1915 | "shards": [ | |
1916 | { | |
1917 | "data_digest": "0x04cfa72f", | |
1918 | "omap_digest": "0xffffffff", | |
1919 | "size": 2048, | |
1920 | "errors": [], | |
1921 | "shard": 2, | |
1922 | "osd": 0 | |
1923 | }, | |
1924 | { | |
1925 | "size": 9, | |
1926 | "errors": [ | |
1927 | "read_error", | |
1928 | "size_mismatch_oi" | |
1929 | ], | |
1930 | "shard": 1, | |
1931 | "osd": 1 | |
1932 | }, | |
1933 | { | |
1934 | "data_digest": "0x04cfa72f", | |
1935 | "omap_digest": "0xffffffff", | |
1936 | "size": 2048, | |
1937 | "errors": [], | |
1938 | "shard": 0, | |
1939 | "osd": 2 | |
1940 | } | |
1941 | ], | |
1942 | "selected_object_info": "2:9175b684:::EOBJ1:head(21'1 client.4179.0:1 dirty|data_digest|omap_digest s 7 uv 1 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1943 | "union_shard_errors": [ | |
1944 | "read_error", | |
1945 | "size_mismatch_oi" | |
1946 | ], | |
1947 | "errors": [ | |
1948 | "size_mismatch" | |
1949 | ], | |
1950 | "object": { | |
1951 | "version": 1, | |
1952 | "snap": "head", | |
1953 | "locator": "", | |
1954 | "nspace": "", | |
1955 | "name": "EOBJ1" | |
1956 | } | |
1957 | }, | |
1958 | { | |
1959 | "shards": [ | |
1960 | { | |
1961 | "size": 2048, | |
1962 | "errors": [ | |
1963 | "ec_hash_error" | |
1964 | ], | |
1965 | "shard": 2, | |
1966 | "osd": 0 | |
1967 | }, | |
1968 | { | |
1969 | "data_digest": "0x04cfa72f", | |
1970 | "omap_digest": "0xffffffff", | |
1971 | "size": 2048, | |
1972 | "errors": [], | |
1973 | "shard": 1, | |
1974 | "osd": 1 | |
1975 | }, | |
1976 | { | |
1977 | "data_digest": "0x04cfa72f", | |
1978 | "omap_digest": "0xffffffff", | |
1979 | "size": 2048, | |
1980 | "errors": [], | |
1981 | "shard": 0, | |
1982 | "osd": 2 | |
1983 | } | |
1984 | ], | |
1985 | "selected_object_info": "2:9babd184:::EOBJ2:head(29'2 client.4217.0:1 dirty|data_digest|omap_digest s 7 uv 2 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
1986 | "union_shard_errors": [ | |
1987 | "ec_hash_error" | |
1988 | ], | |
1989 | "errors": [], | |
1990 | "object": { | |
1991 | "version": 2, | |
1992 | "snap": "head", | |
1993 | "locator": "", | |
1994 | "nspace": "", | |
1995 | "name": "EOBJ2" | |
1996 | } | |
1997 | }, | |
1998 | { | |
1999 | "shards": [ | |
2000 | { | |
2001 | "data_digest": "0x04cfa72f", | |
2002 | "omap_digest": "0xffffffff", | |
2003 | "size": 2048, | |
2004 | "errors": [], | |
2005 | "shard": 2, | |
2006 | "osd": 0 | |
2007 | }, | |
2008 | { | |
2009 | "errors": [ | |
2010 | "missing" | |
2011 | ], | |
2012 | "shard": 1, | |
2013 | "osd": 1 | |
2014 | }, | |
2015 | { | |
2016 | "data_digest": "0x04cfa72f", | |
2017 | "omap_digest": "0xffffffff", | |
2018 | "size": 2048, | |
2019 | "errors": [], | |
2020 | "shard": 0, | |
2021 | "osd": 2 | |
2022 | } | |
2023 | ], | |
2024 | "selected_object_info": "2:b197b25d:::EOBJ3:head(37'3 client.4251.0:1 dirty|data_digest|omap_digest s 7 uv 3 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
2025 | "union_shard_errors": [ | |
2026 | "missing" | |
2027 | ], | |
2028 | "errors": [], | |
2029 | "object": { | |
2030 | "version": 3, | |
2031 | "snap": "head", | |
2032 | "locator": "", | |
2033 | "nspace": "", | |
2034 | "name": "EOBJ3" | |
2035 | } | |
2036 | }, | |
2037 | { | |
2038 | "shards": [ | |
2039 | { | |
2040 | "attrs": [ | |
2041 | { | |
2042 | "Base64": true, | |
2043 | "value": "", | |
2044 | "name": "_" | |
2045 | }, | |
2046 | { | |
2047 | "Base64": false, | |
2048 | "value": "bad-val", | |
2049 | "name": "_key1-EOBJ4" | |
2050 | }, | |
2051 | { | |
2052 | "Base64": false, | |
2053 | "value": "val3-EOBJ4", | |
2054 | "name": "_key3-EOBJ4" | |
2055 | }, | |
2056 | { | |
2057 | "Base64": true, | |
2058 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
2059 | "name": "hinfo_key" | |
2060 | }, | |
2061 | { | |
2062 | "Base64": true, | |
2063 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
2064 | "name": "snapset" | |
2065 | } | |
2066 | ], | |
2067 | "data_digest": "0x04cfa72f", | |
2068 | "omap_digest": "0xffffffff", | |
2069 | "size": 2048, | |
2070 | "errors": [], | |
2071 | "shard": 2, | |
2072 | "osd": 0 | |
2073 | }, | |
2074 | { | |
2075 | "attrs": [ | |
2076 | { | |
2077 | "Base64": true, | |
2078 | "value": "", | |
2079 | "name": "_" | |
2080 | }, | |
2081 | { | |
2082 | "Base64": false, | |
2083 | "value": "val1-EOBJ4", | |
2084 | "name": "_key1-EOBJ4" | |
2085 | }, | |
2086 | { | |
2087 | "Base64": false, | |
2088 | "value": "val2-EOBJ4", | |
2089 | "name": "_key2-EOBJ4" | |
2090 | }, | |
2091 | { | |
2092 | "Base64": true, | |
2093 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
2094 | "name": "hinfo_key" | |
2095 | }, | |
2096 | { | |
2097 | "Base64": true, | |
2098 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
2099 | "name": "snapset" | |
2100 | } | |
2101 | ], | |
2102 | "data_digest": "0x04cfa72f", | |
2103 | "omap_digest": "0xffffffff", | |
2104 | "size": 2048, | |
2105 | "errors": [], | |
2106 | "shard": 1, | |
2107 | "osd": 1 | |
2108 | }, | |
2109 | { | |
2110 | "attrs": [ | |
2111 | { | |
2112 | "Base64": true, | |
2113 | "value": "", | |
2114 | "name": "_" | |
2115 | }, | |
2116 | { | |
2117 | "Base64": false, | |
2118 | "value": "val1-EOBJ4", | |
2119 | "name": "_key1-EOBJ4" | |
2120 | }, | |
2121 | { | |
2122 | "Base64": false, | |
2123 | "value": "val2-EOBJ4", | |
2124 | "name": "_key2-EOBJ4" | |
2125 | }, | |
2126 | { | |
2127 | "Base64": true, | |
2128 | "value": "AQEYAAAAAAgAAAAAAAADAAAAL6fPBLB8dlsvp88E", | |
2129 | "name": "hinfo_key" | |
2130 | }, | |
2131 | { | |
2132 | "Base64": true, | |
2133 | "value": "AgIZAAAAAAAAAAAAAAABAAAAAAAAAAAAAAAAAAAAAA==", | |
2134 | "name": "snapset" | |
2135 | } | |
2136 | ], | |
2137 | "data_digest": "0x04cfa72f", | |
2138 | "omap_digest": "0xffffffff", | |
2139 | "size": 2048, | |
2140 | "errors": [], | |
2141 | "shard": 0, | |
2142 | "osd": 2 | |
2143 | } | |
2144 | ], | |
2145 | "selected_object_info": "2:5e723e06:::EOBJ4:head(45'6 client.4289.0:1 dirty|data_digest|omap_digest s 7 uv 6 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
2146 | "union_shard_errors": [], | |
2147 | "errors": [ | |
2148 | "attr_value_mismatch", | |
2149 | "attr_name_mismatch" | |
2150 | ], | |
2151 | "object": { | |
2152 | "version": 6, | |
2153 | "snap": "head", | |
2154 | "locator": "", | |
2155 | "nspace": "", | |
2156 | "name": "EOBJ4" | |
2157 | } | |
2158 | }, | |
2159 | { | |
2160 | "shards": [ | |
2161 | { | |
2162 | "data_digest": "0x04cfa72f", | |
2163 | "omap_digest": "0xffffffff", | |
2164 | "size": 2048, | |
2165 | "errors": [], | |
2166 | "shard": 2, | |
2167 | "osd": 0 | |
2168 | }, | |
2169 | { | |
2170 | "size": 4096, | |
2171 | "errors": [ | |
2172 | "size_mismatch_oi", | |
2173 | "ec_size_error" | |
2174 | ], | |
2175 | "shard": 1, | |
2176 | "osd": 1 | |
2177 | }, | |
2178 | { | |
2179 | "data_digest": "0x04cfa72f", | |
2180 | "omap_digest": "0xffffffff", | |
2181 | "size": 2048, | |
2182 | "errors": [], | |
2183 | "shard": 0, | |
2184 | "osd": 2 | |
2185 | } | |
2186 | ], | |
2187 | "selected_object_info": "2:8549dfb5:::EOBJ5:head(65'7 client.4441.0:1 dirty|data_digest|omap_digest s 7 uv 7 dd 2ddbf8f5 od ffffffff alloc_hint [0 0 0])", | |
2188 | "union_shard_errors": [ | |
2189 | "size_mismatch_oi", | |
2190 | "ec_size_error" | |
2191 | ], | |
2192 | "errors": [ | |
2193 | "size_mismatch" | |
2194 | ], | |
2195 | "object": { | |
2196 | "version": 7, | |
2197 | "snap": "head", | |
2198 | "locator": "", | |
2199 | "nspace": "", | |
2200 | "name": "EOBJ5" | |
2201 | } | |
2202 | } | |
2203 | ], | |
2204 | "epoch": 0 | |
2205 | } | |
2206 | EOF | |
2207 | ||
2208 | jq "$jqfilter" $dir/json | python -c "$sortkeys" | sed -e "$sedfilter" > $dir/csjson | |
2209 | if [ "$allow_overwrites" = "true" ] | |
2210 | then | |
2211 | grep -v data_digest $dir/csjson | grep -v ec_size_error > $dir/csjson | |
2212 | grep -v data_digest $dir/checkcsjson | grep -v ec_size_error > $dir/checkcsjson | |
2213 | fi | |
2214 | diff -y $termwidth $dir/checkcsjson $dir/csjson || test $getjson = "yes" || return 1 | |
2215 | if test $getjson = "yes" | |
2216 | then | |
2217 | jq '.' $dir/json > save4.json | |
2218 | fi | |
2219 | ||
2220 | if which jsonschema > /dev/null; | |
2221 | then | |
2222 | jsonschema -i $dir/json $CEPH_ROOT/doc/rados/command/list-inconsistent-obj.json || return 1 | |
2223 | fi | |
2224 | ||
2225 | rados rmpool $poolname $poolname --yes-i-really-really-mean-it | |
2226 | teardown $dir || return 1 | |
2227 | } | |
2228 | ||
2229 | function TEST_corrupt_scrub_erasure_appends() { | |
2230 | corrupt_scrub_erasure $1 false | |
2231 | } | |
2232 | ||
2233 | function TEST_corrupt_scrub_erasure_overwrites() { | |
2234 | corrupt_scrub_erasure $1 true | |
2235 | } | |
2236 | ||
2237 | # | |
2238 | # Test to make sure that a periodic scrub won't cause deep-scrub info to be lost | |
2239 | # | |
2240 | function TEST_periodic_scrub_replicated() { | |
2241 | local dir=$1 | |
2242 | local poolname=psr_pool | |
2243 | local objname=POBJ | |
2244 | ||
2245 | setup $dir || return 1 | |
2246 | run_mon $dir a --osd_pool_default_size=2 || return 1 | |
2247 | run_mgr $dir x || return 1 | |
2248 | local ceph_osd_args="--osd-scrub-interval-randomize-ratio=0 --osd-deep-scrub-randomize-ratio=0" | |
2249 | run_osd $dir 0 $ceph_osd_args || return 1 | |
2250 | run_osd $dir 1 $ceph_osd_args || return 1 | |
2251 | wait_for_clean || return 1 | |
2252 | ||
2253 | ceph osd pool create $poolname 1 1 || return 1 | |
2254 | wait_for_clean || return 1 | |
2255 | ||
2256 | local osd=0 | |
2257 | add_something $dir $poolname $objname scrub || return 1 | |
2258 | local primary=$(get_primary $poolname $objname) | |
2259 | local pg=$(get_pg $poolname $objname) | |
2260 | ||
2261 | # Add deep-scrub only error | |
2262 | local payload=UVWXYZ | |
2263 | echo $payload > $dir/CORRUPT | |
2264 | # Uses $ceph_osd_args for osd restart | |
2265 | objectstore_tool $dir $osd $objname set-bytes $dir/CORRUPT || return 1 | |
2266 | ||
2267 | # No scrub information available, so expect failure | |
2268 | set -o pipefail | |
2269 | ! rados list-inconsistent-obj $pg | jq '.' || return 1 | |
2270 | set +o pipefail | |
2271 | ||
2272 | pg_deep_scrub $pg || return 1 | |
2273 | ||
2274 | # Make sure bad object found | |
2275 | rados list-inconsistent-obj $pg | jq '.' | grep -q $objname || return 1 | |
2276 | ||
2277 | local last_scrub=$(get_last_scrub_stamp $pg) | |
2278 | # Fake a schedule scrub | |
2279 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.${primary}.asok \ | |
2280 | trigger_scrub $pg || return 1 | |
2281 | # Wait for schedule regular scrub | |
2282 | wait_for_scrub $pg "$last_scrub" | |
2283 | ||
2284 | # It needed to be upgraded | |
2285 | grep -q "Deep scrub errors, upgrading scrub to deep-scrub" $dir/osd.${primary}.log || return 1 | |
2286 | ||
2287 | # Bad object still known | |
2288 | rados list-inconsistent-obj $pg | jq '.' | grep -q $objname || return 1 | |
2289 | ||
2290 | # Can't upgrade with this set | |
2291 | ceph osd set nodeep-scrub | |
2292 | ||
2293 | # Fake a schedule scrub | |
2294 | local last_scrub=$(get_last_scrub_stamp $pg) | |
2295 | CEPH_ARGS='' ceph --admin-daemon $dir/ceph-osd.${primary}.asok \ | |
2296 | trigger_scrub $pg || return 1 | |
2297 | # Wait for schedule regular scrub | |
2298 | # to notice scrub and skip it | |
2299 | local found=false | |
2300 | for i in $(seq 14 -1 0) | |
2301 | do | |
2302 | sleep 1 | |
2303 | ! grep -q "Regular scrub skipped due to deep-scrub errors and nodeep-scrub set" $dir/osd.${primary}.log || { found=true ; break; } | |
2304 | echo Time left: $i seconds | |
2305 | done | |
2306 | test $found = "true" || return 1 | |
2307 | ||
2308 | # Bad object still known | |
2309 | rados list-inconsistent-obj $pg | jq '.' | grep -q $objname || return 1 | |
2310 | ||
2311 | # Request a regular scrub and it will be done | |
2312 | local scrub_backoff_ratio=$(get_config osd ${primary} osd_scrub_backoff_ratio) | |
2313 | set_config osd ${primary} osd_scrub_backoff_ratio 0 | |
2314 | pg_scrub $pg | |
2315 | sleep 1 | |
2316 | set_config osd ${primary} osd_scrub_backoff_ratio $scrub_backoff_ratio | |
2317 | grep -q "Regular scrub request, losing deep-scrub details" $dir/osd.${primary}.log || return 1 | |
2318 | ||
2319 | # deep-scrub error is no longer present | |
2320 | rados list-inconsistent-obj $pg | jq '.' | grep -qv $objname || return 1 | |
2321 | } | |
2322 | ||
2323 | ||
2324 | main osd-scrub-repair "$@" | |
2325 | ||
2326 | # Local Variables: | |
2327 | # compile-command: "cd ../.. ; make -j4 && \ | |
2328 | # test/osd/osd-scrub-repair.sh # TEST_corrupt_and_repair_replicated" | |
2329 | # End: |