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
|
#!/usr/bin/env bash
# version 2023-01-11
# IMPORTANT: affects order produced by 'sort' and 'ceph-diff-sorted'
# relies on this ordering
export LC_ALL=C
# If your ceph.conf is not in /etc/ceph, then set CEPH_CONF="-c /path/to/ceph.conf"
trap "exit 1" TERM
TOP_PID=$$
out_dir="."
timestamp=$(date -u +%Y%m%d%H%M%S)
lspools_err="${out_dir}/lspools-${timestamp}.error"
rados_out="${out_dir}/rados-${timestamp}.intermediate"
rados_odd="${out_dir}/rados-${timestamp}.issues"
rados_err="${out_dir}/rados-${timestamp}.error"
rgwadmin_out="${out_dir}/radosgw-admin-${timestamp}.intermediate"
rgwadmin_err="${out_dir}/radosgw-admin-${timestamp}.error"
delta_out="${out_dir}/orphan-list-${timestamp}.out"
log() {
echo $(date +%F\ %T) $(hostname -s) "$1"
}
usage() {
>&2 cat << EOF
Usage: $0 [-h] "<radospools>" [<temp_dir>]
Where:
-h This help output
<radospools> The RGW data pool name, if omitted, pool name will be
prompted for during execution.
If specifying multiple pools, please use space separated
list and wrap the entire list in quotes.
<temp_dir> Optionally, set the directory to use for temp space.
This may be required if /tmp is low on space.
NOTES:
- This tool should be ran on a node with ceph-radosgw package installed.
Specifically, it needs the 'ceph-diff-tool' command from that package.
- This tool is currently considered to be EXPERIMENTAL.
- False positives are possible. False positives would likely
appear as objects that were never deleted and are fully
intact. All results should therefore be verified.
WARNING:
- Indexless buckets will appear as 100% orphan objects.
- Therefore, this tool MUST NOT be used in environments with indexless
buckets.
EOF
exit 1
}
#
# checkReturn RETURNCODE MESSAGE TERMINATE
# RETURNCODE - ( usually $? ) of previous command
# MESSAGE - Message to print on non-zero return code
# TERMINATE - non-empty == terminate the script on non-zero return code
#
checkReturn() {
if [ $1 -ne 0 ]; then
error_addon=""
if [ ! -z "$3" ]; then
error_addon="; Terminating"
fi
log "ERROR: ${2} failed: returned ${1}${error_addon}"
if [ ! -z "$3" ]; then
>&2 echo
>&2 echo '***'
>&2 echo '*** WARNING: The results are incomplete. Do not use! ***'
>&2 echo '***'
kill -s TERM $TOP_PID
fi
fi
}
prompt_pool() {
# note: all prompts go to stderr so stdout contains just the result
>&2 echo "Available pools:"
rados ${CEPH_CONF} lspools >"$temp_file" 2>"$lspools_err"
checkReturn $? "Listing pools failed" 1
>&2 sed 's/^/ /' "$temp_file" # list pools and indent
>&2 printf "Which pool do you want to search for orphans (for multiple, use space-separated list)? "
local mypool
read mypool
echo $mypool
}
radosgw_radoslist() {
log "Running 'radosgw-admin bucket radoslist'."
rm -f "$rgwadmin_flag" &> /dev/null
radosgw-admin ${CEPH_CONF} bucket radoslist >"$rgwadmin_out" 2>"$rgwadmin_err"
RETVAL=$?
if [ "$RETVAL" -ne 0 ] ;then
touch "$rgwadmin_flag"
fi
checkReturn $RETVAL "radosgw-admin radoslist" 1
log "Completed 'radosgw-admin bucket radoslist'."
log "Sorting 'radosgw-admin bucket radoslist' output."
sort -T ${temp_prefix} -u "$rgwadmin_out" > "$rgwadmin_temp"
checkReturn $? "Sorting 'radosgw-admin bucket radoslist' output" 1
log "Completed sorting 'radosgw-admin bucket radoslist'."
log "Moving 'radosgw-admin bucket radoslist' output."
mv -f "$rgwadmin_temp" "$rgwadmin_out"
checkReturn $? "Moving 'radosgw-admin bucket radoslist' output" 1
log "Completed moving 'radosgw-admin bucket radoslist' output."
}
rados_ls() {
log "Starting 'rados ls' function."
rm -f "$rados_flag" &> /dev/null
rm -f "$rados_out" &> /dev/null
local mypool
for mypool in $pool; do
log "Running 'rados ls' on pool ${mypool}."
rados ${CEPH_CONF} ls --pool="$mypool" --all >>"$rados_out" 2>"$rados_err"
RETVAL=$?
if [ "$RETVAL" -ne 0 ] ;then
touch "$rados_flag"
fi
checkReturn $RETVAL "'rados ls' on pool ${mypool}" 1
log "Completed 'rados ls' on pool ${mypool}."
done
if [ ! -e "$rados_flag" ]; then
# NOTE: Each entry (line of output) of `rados ls --all` should be in
# one of four formats depending on whether or not an entry has a
# namespace and/or locator:
#
# <TAB>oid
# <TAB>oid<TAB>locator
# namespace<TAB>oid
# namespace<TAB>oid<TAB>locator
#
# Any occurrences of the 2nd, 3rd, or 4th (i.e., existence of
# namespace and/or locator) should cause the create of the "odd" file
# and an explanation in the output, and those entries will not be
# retained, and therefore they will not be called out as orphans. They
# will need special handling by the end-user as we do not expect
# namespaces or locators.
# check for namespaces -- any line that does not begin with a tab
# indicates a namespace; add those to "odd" file and set flag; note:
# this also picks up entries with namespace and locator
log "Checking for namespaces"
grep --text $'^[^\t]' "$rados_out" >"$rados_odd"
if [ "${PIPESTATUS[0]}" -eq 0 ] ;then
log "Namespaces found"
namespace_found=1
fi
# check for locators (w/o namespace); we idenitfy them by skipping
# past the empty namespace (i.e., one TAB), skipping past the oid,
# then looking for a TAB; note we use egrep to get the '+' character
# and the $ in front of the ' allows the \t to be interpreted as a TAB
log "Checking for locators"
egrep --text $'^\t[[:graph:]]+\t' "$rados_out" >>"$rados_odd"
if [ "${PIPESTATUS[0]}" -eq 0 ] ;then
log "Locator found"
locator_found=1
fi
# extract the entries that are just oids (i.e., no namespace or
# locator) for further processing; only look at lines that begin with
# a TAB and do not contain a second TAB, and then grab everything
# after the initial TAB
log "Generating final 'rados ls' output (without namespaces or locators)"
grep --text $'^\t' "$rados_out" | grep --text -v $'^\t.*\t' | sed -E 's/^\t//' >"$temp_file"
mv -f "$temp_file" "$rados_out"
log "Sorting 'rados ls' output(s)."
sort -T ${temp_prefix} -u "$rados_out" >"$temp_file"
checkReturn $? "Sorting 'rados ls' output(s)" 1
log "Sorting 'rados ls' output(s) complete."
log "Moving sorted output(s)."
mv -f "$temp_file" "$rados_out"
checkReturn $? "Moving temp file to output file" 1
fi
}
temp_prefix="/tmp"
if [ ! -z "$2" ]; then
if [ -d "$2" ]; then
temp_prefix=$2
else
echo
echo "ERROR: Provided temp directory does not exist: ${2}"
usage
fi
temp_prefix="$2"
fi
temp_file=${temp_prefix}/temp.$$
rados_flag=${temp_prefix}/rados_flag.$$
rgwadmin_flag=${temp_prefix}/rgwadmin_flag.$$
rgwadmin_temp=${temp_prefix}/rgwadmin_temp.$$
if [ $# -eq 0 ] ;then
pool="$(prompt_pool)"
else
if [ "$1" == "-h" ]; then
usage
fi
pool="$1"
fi
error=0
rados ${CEPH_CONF} lspools > $temp_file
for mypool in $pool; do
if [ $(grep -c "^${mypool}$" "${temp_file}") -eq 0 ]; then
echo
echo "ERROR: Supplied pool does not exist: ${mypool}"
error=1
fi
done
if [ $error -gt 0 ]; then
echo "Terminating"
exit 1
fi
log "Pool is \"$pool\"."
log "Note: output files produced will be tagged with the current timestamp -- ${timestamp}."
rados_ls
radosgw_radoslist
#
# Check for any empty output files
#
for myfile in $rados_out $rgwadmin_out; do
if [ ! -s "${myfile}" ]; then
log "ERROR: Empty file detected: ${myfile}"
log "ERROR: RESULTS ARE INCOMPLETE - DO NOT USE"
exit 1
fi
done
log "Computing delta..."
ceph-diff-sorted "$rados_out" "$rgwadmin_out" | grep --text "^<" | sed 's/^< *//' >"$delta_out"
# use PIPESTATUS to get at exit status of first process in above pipe;
# 0 means same, 1 means different, >1 means error
if [ "${PIPESTATUS[0]}" -gt 1 ] ;then
log "ERROR: ceph-diff-sorted failed with status: ${PIPESTATUS[0]}"
log "TERMINATING - Results are incomplete - DO NOT USE"
exit 1
fi
log "Computing results..."
found=$(wc -l < "$delta_out")
possible=$(wc -l < "$rados_out")
percentage=0
if [ $possible -ne 0 ] ;then
percentage=$(expr 100 \* $found / $possible)
fi
echo "$found potential orphans found out of a possible $possible (${percentage}%)."
echo "The results can be found in '${delta_out}'."
echo " Intermediate files are '${rados_out}' and '${rgwadmin_out}'."
if [ -n "$namespace_found" -o -n "$locator_found" ] ;then
echo " Note: 'rados ls' found entries that might be in a namespace or might"
echo " have a locator; see '${rados_odd}' for those entries."
fi
echo "***"
echo "*** WARNING: This is EXPERIMENTAL code and the results should be used"
echo "*** only with CAUTION!"
echo "***"
echo "Done at $(date +%F\ %T)."
|