Skip to content

Pt stalk mongodb #18

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
fipar wants to merge 3 commits into from
Closed

Pt stalk mongodb #18

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3cd77a6
added skel for mongo support, including suppresion of mysql collectio…
fipar Jun 11, 2015
f8a61cb
added trg_mongo_default and made it the default function when --mongo…
fipar Jun 11, 2015
10b6232
Changed:
fipar Jun 11, 2015
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
201 changes: 116 additions & 85 deletions bin/pt-stalk
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -786,61 +786,75 @@ CMD_SYSCTL="${CMD_SYSCTL:-"$(_which sysctl)"}"
CMD_TCPDUMP="${CMD_TCPDUMP:-"$(_which tcpdump)"}"
CMD_VMSTAT="${CMD_VMSTAT:-"$(_which vmstat)"}"
CMD_DMESG="${CMD_DMESG:-"$(_which dmesg)"}"
CMD_MONGO="${CMD_MONGO:-"$(_which mongo)"}"

[ -z "$CMD_SYSCTL" -a -x "/sbin/sysctl" ] && CMD_SYSCTL="/sbin/sysctl"

collect_mongo() {
local n=$1
$CMD_MONGO $EXT_ARGV --eval 'printjson(db.currentOp(true))' >> "$d/$p-currentOp$n"
$CMD_MONGO $EXT_ARGV --eval 'printjson(db.isMaster())' >> "$d/$p-isMaster$n"
$CMD_MONGO $EXT_ARGV --eval 'printjson(sh.status())' >> "$d/$p-shStatus$n"
$CMD_MONGO $EXT_ARGV --eval 'printjson(rs.status())' >> "$d/$p-rsStatus$n"
$CMD_MONGO $EXT_ARGV --eval 'printjson(db.serverStatus())' >> "$d/$p-serverStatus$n"
[ $n -eq 2 ] && $CMD_MONGO $EXT_ARGV --eval 'db.adminCommand({getLog:"*"})["names"].forEach(function (e,a,i){ db.adminCommand({getLog:e})["log"].forEach(function (e,a,i){print(e)})})' >> "$d/$p-logs" # We don't need to run this twice since it will be mostly duplicate lines, so we only run it the second time, which gives us the most log content
}

collect() {
local d="$1" # directory to save results in
local p="$2" # prefix for each result file

local mysqld_pid=$(_pidof mysqld | awk '{print $1; exit;}')

if [ "$CMD_PMAP" -a "$mysqld_pid" ]; then
if $CMD_PMAP --help 2>&1 | grep -- -x >/dev/null 2>&1 ; then
$CMD_PMAP -x $mysqld_pid > "$d/$p-pmap"
else
$CMD_PMAP $mysqld_pid > "$d/$p-pmap"
fi
fi

if [ "$CMD_GDB" -a "$OPT_COLLECT_GDB" -a "$mysqld_pid" ]; then
$CMD_GDB \
-ex "set pagination 0" \
-ex "thread apply all bt" \
--batch -p $mysqld_pid \
>> "$d/$p-stacktrace"
fi

$CMD_MYSQL $EXT_ARGV -e 'SHOW GLOBAL VARIABLES' >> "$d/$p-variables" &
sleep .2

local mysql_version="$(awk '/^version[^_]/{print substr($2,1,3)}' "$d/$p-variables")"

local mysql_error_log="$(awk '/log_error/{print $2}' "$d/$p-variables")"
if [ -z "$mysql_error_log" -a "$mysqld_pid" ]; then
mysql_error_log="$(ls -l /proc/$mysqld_pid/fd | awk '/ 2 ->/{print $NF}')"
fi

local tail_error_log_pid=""
if [ "$mysql_error_log" ]; then
log "The MySQL error log seems to be $mysql_error_log"
tail -f "$mysql_error_log" >"$d/$p-log_error" &
tail_error_log_pid=$!

$CMD_MYSQLADMIN $EXT_ARGV debug
else
log "Could not find the MySQL error log"
fi

if [ "${mysql_version}" '>' "5.1" ]; then
local mutex="SHOW ENGINE INNODB MUTEX"
else
local mutex="SHOW MUTEX STATUS"
local mysqld_pid=""
[ "$OPT_MONGO" == "yes" ] || local mysqld_pid=$(_pidof mysqld | awk '{print $1; exit;}')

if [ "$OPT_MONGO" != "yes" ]; then
if [ "$CMD_PMAP" -a "$mysqld_pid" ]; then
if $CMD_PMAP --help 2>&1 | grep -- -x >/dev/null 2>&1 ; then
$CMD_PMAP -x $mysqld_pid > "$d/$p-pmap"
else
# Some pmap's apparently don't support -x (issue 116).
$CMD_PMAP $mysqld_pid > "$d/$p-pmap"
fi
fi

if [ "$CMD_GDB" -a "$OPT_COLLECT_GDB" -a "$mysqld_pid" ]; then
$CMD_GDB \
-ex "set pagination 0" \
-ex "thread apply all bt" \
--batch -p $mysqld_pid \
>> "$d/$p-stacktrace"
fi

$CMD_MYSQL $EXT_ARGV -e 'SHOW GLOBAL VARIABLES' >> "$d/$p-variables" &
sleep .2

local mysql_version="$(awk '/^version[^_]/{print substr($2,1,3)}' "$d/$p-variables")"

local mysql_error_log="$(awk '/log_error/{print $2}' "$d/$p-variables")"
if [ -z "$mysql_error_log" -a "$mysqld_pid" ]; then
mysql_error_log="$(ls -l /proc/$mysqld_pid/fd | awk '/ 2 ->/{print $NF}')"
fi

local tail_error_log_pid=""
if [ "$mysql_error_log" ]; then
log "The MySQL error log seems to be $mysql_error_log"
tail -f "$mysql_error_log" >"$d/$p-log_error" &
tail_error_log_pid=$!

$CMD_MYSQLADMIN $EXT_ARGV debug
else
log "Could not find the MySQL error log"
fi

if [ "${mysql_version}" '>' "5.1" ]; then
local mutex="SHOW ENGINE INNODB MUTEX"
else
local mutex="SHOW MUTEX STATUS"
fi
innodb_status 1
$CMD_MYSQL $EXT_ARGV -e "$mutex" >> "$d/$p-mutex-status1" &
open_tables >> "$d/$p-opentables1" &
fi
innodb_status 1
tokudb_status 1
$CMD_MYSQL $EXT_ARGV -e "$mutex" >> "$d/$p-mutex-status1" &
open_tables >> "$d/$p-opentables1" &

local tcpdump_pid=""
if [ "$CMD_TCPDUMP" -a "$OPT_COLLECT_TCPDUMP" ]; then
Expand All @@ -858,7 +872,7 @@ collect() {
have_oprofile="yes"
fi
elif [ "$CMD_STRACE" -a "$OPT_COLLECT_STRACE" -a "$mysqld_pid" ]; then
$CMD_STRACE -T -s 0 -f -p $mysqld_pid -o "$d/$p-strace" &
$CMD_STRACE -T -s 0 -f -p $mysqld_pid > "${DEST}/$d-strace" &
local strace_pid=$!
fi

Expand All @@ -871,7 +885,6 @@ collect() {
$CMD_SYSCTL -a >> "$d/$p-sysctl" &
fi

# collect dmesg events from 60 seconds ago until present
if [ "$CMD_DMESG" ]; then
local UPTIME=`cat /proc/uptime | awk '{ print $1 }'`
local START_TIME=$(echo "$UPTIME 60" | awk '{print ($1 - $2)}')
Expand All @@ -892,17 +905,19 @@ collect() {
$CMD_MPSTAT -P ALL $OPT_RUN_TIME 1 >> "$d/$p-mpstat-overall" &
fi

if [ "$OPT_MONGO" != "yes" ]; then
$CMD_MYSQLADMIN $EXT_ARGV ext -i$OPT_SLEEP_COLLECT -c$cnt >>"$d/$p-mysqladmin" &
local mysqladmin_pid=$!

$CMD_MYSQLADMIN $EXT_ARGV ext -i$OPT_SLEEP_COLLECT -c$cnt >>"$d/$p-mysqladmin" &
local mysqladmin_pid=$!

local have_lock_waits_table=""
$CMD_MYSQL $EXT_ARGV -e "SHOW TABLES FROM INFORMATION_SCHEMA" \
| grep -i "INNODB_LOCK_WAITS" >/dev/null 2>&1
if [ $? -eq 0 ]; then
have_lock_waits_table="yes"
local have_lock_waits_table=""
$CMD_MYSQL $EXT_ARGV -e "SHOW TABLES FROM INFORMATION_SCHEMA" \
| grep -i "INNODB_LOCK_WAITS" >/dev/null 2>&1
if [ $? -eq 0 ]; then
have_lock_waits_table="yes"
fi
fi


[ "$OPT_MONGO" == "yes" ] && collect_mongo 1 &
log "Loop start: $(date +'TS %s.%N %F %T')"
local start_time=$(date +'%s')
local curr_time=$start_time
Expand Down Expand Up @@ -941,11 +956,13 @@ collect() {
(echo $ts; df -k) >> "$d/$p-df" &
(echo $ts; netstat -antp) >> "$d/$p-netstat" &
(echo $ts; netstat -s) >> "$d/$p-netstat_s" &
(echo $ts; $CMD_MYSQL $EXT_ARGV -e "SHOW FULL PROCESSLIST\G") \
>> "$d/$p-processlist" &
if [ "$have_lock_waits_table" ]; then
(echo $ts; lock_waits) >>"$d/$p-lock-waits" &
(echo $ts; transactions) >>"$d/$p-transactions" &
if [ "$OPT_MONGO" != "yes" ]; then
(echo $ts; $CMD_MYSQL $EXT_ARGV -e "SHOW FULL PROCESSLIST\G") \
>> "$d/$p-processlist" &
if [ "$have_lock_waits_table" ]; then
(echo $ts; lock_waits) >>"$d/$p-lock-waits" &
(echo $ts; transactions) >>"$d/$p-transactions" &
fi
fi

curr_time=$(date +'%s')
Expand Down Expand Up @@ -987,22 +1004,25 @@ collect() {
kill -s 15 $strace_pid
[ "$mysqld_pid" ] && kill -s 18 $mysqld_pid
fi

if [ "$OPT_MONGO" != "yes" ]; then
innodb_status 2
$CMD_MYSQL $EXT_ARGV -e "$mutex" >> "$d/$p-mutex-status2" &
open_tables >> "$d/$p-opentables2" &

innodb_status 2
tokudb_status 2
$CMD_MYSQL $EXT_ARGV -e "$mutex" >> "$d/$p-mutex-status2" &
open_tables >> "$d/$p-opentables2" &

kill $mysqladmin_pid
[ "$tail_error_log_pid" ] && kill $tail_error_log_pid
kill $mysqladmin_pid
[ "$tail_error_log_pid" ] && kill $tail_error_log_pid
else
collect_mongo 2 &
fi
[ "$tcpdump_pid" ] && kill $tcpdump_pid

hostname > "$d/$p-hostname"

wait_for_subshells $OPT_RUN_TIME
kill_all_subshells
for file in "$d/$p-"*; do
if [ -z "$(grep -v '^TS ' --max-count 10 "$file")" ]; then
if [ -z "$(grep -v '^TS ' --max-count 1 "$file")" ]; then
log "Removing empty file $file";
rm "$file"
fi
Expand All @@ -1019,7 +1039,7 @@ open_tables() {
}

lock_waits() {
local sql1="SELECT SQL_NO_CACHE
local sql1="SELECT
CONCAT('thread ', b.trx_mysql_thread_id, ' from ', p.host) AS who_blocks,
IF(p.command = \"Sleep\", p.time, 0) AS idle_in_trx,
MAX(TIMESTAMPDIFF(SECOND, r.trx_wait_started, CURRENT_TIMESTAMP)) AS max_wait_time,
Expand All @@ -1031,7 +1051,7 @@ lock_waits() {
GROUP BY who_blocks ORDER BY num_waiters DESC\G"
$CMD_MYSQL $EXT_ARGV -e "$sql1"

local sql2="SELECT SQL_NO_CACHE
local sql2="SELECT
r.trx_id AS waiting_trx_id,
r.trx_mysql_thread_id AS waiting_thread,
TIMESTAMPDIFF(SECOND, r.trx_wait_started, CURRENT_TIMESTAMP) AS wait_time,
Expand All @@ -1052,16 +1072,9 @@ lock_waits() {
}

transactions() {
$CMD_MYSQL $EXT_ARGV -e "SELECT SQL_NO_CACHE * FROM INFORMATION_SCHEMA.INNODB_TRX\G"
$CMD_MYSQL $EXT_ARGV -e "SELECT SQL_NO_CACHE * FROM INFORMATION_SCHEMA.INNODB_LOCKS\G"
$CMD_MYSQL $EXT_ARGV -e "SELECT SQL_NO_CACHE * FROM INFORMATION_SCHEMA.INNODB_LOCK_WAITS\G"
}

tokudb_status() {
local n=$1

$CMD_MYSQL $EXT_ARGV -e "SHOW ENGINE TOKUDB STATUS\G" \
>> "$d/$p-tokudbstatus$n" || rm -f "$d/$p-tokudbstatus$n"
$CMD_MYSQL $EXT_ARGV -e "SELECT * FROM INFORMATION_SCHEMA.INNODB_TRX\G"
$CMD_MYSQL $EXT_ARGV -e "SELECT * FROM INFORMATION_SCHEMA.INNODB_LOCKS\G"
$CMD_MYSQL $EXT_ARGV -e "SELECT * FROM INFORMATION_SCHEMA.INNODB_LOCK_WAITS\G"
}

innodb_status() {
Expand Down Expand Up @@ -1098,6 +1111,9 @@ EXIT_REASON=""
TOOL="pt-stalk"
OKTORUN=1
ITER=1
DARWIN=

[ "$(uname)" == "Darwin" ] && DARWIN=yes

# ###########################################################################
# Plugin hooks
Expand Down Expand Up @@ -1183,6 +1199,11 @@ set_trg_func() {
TRIGGER_FUNCTION="trg_plugin"
return 0 # success
else
# If --mongo is used, we only have one built-in function
if [ "$OPT_MONGO" == "yes" ]; then
TRIGGER_FUNCTION="trg_mongo_default"
return 0 # success
fi
# Trigger function is name of a built-in function.
func=$(echo "$func" | tr '[:upper:]' '[:lower:]')
if [ "$func" = "status" -o "$func" = "processlist" ]; then
Expand All @@ -1193,6 +1214,10 @@ set_trg_func() {
return 1 # error
}

trg_mongo_default() {
$CMD_MONGO --quiet --eval "db.currentOp()['inprog'].length"
}

trg_status() {
local var="$1"
mysqladmin $EXT_ARGV extended-status \
Expand Down Expand Up @@ -1237,13 +1262,15 @@ purge_samples() {
local retention_time="$2"

# Delete collect files which more than --retention-time days old.
find "$dir" -warn -type f -mtime +$retention_time -exec rm -f '{}' \;
warn="-warn"
[ -n "$DARWIN" ] && warn=""
find "$dir" $warn -type f -mtime +$retention_time -exec rm -f '{}' \;

local oprofile_dir="/var/lib/oprofile/samples"
if [ -d "$oprofile_dir" ]; then
# "pt_collect_" here needs to match $CMD_OPCONTROL --save=pt_collect_$p
# in collect(). TODO: fix this
find "$oprofile_dir" -warn -depth -type d -name 'pt_collect_*' \
find "$oprofile_dir" $warn -depth -type d -name 'pt_collect_*' \
-mtime +$retention_time -exec rm -rf '{}' \;
fi
}
Expand Down Expand Up @@ -1891,6 +1918,10 @@ type: string
The pattern to use when watching SHOW PROCESSLIST. See L<"--function">
for details.

=item --mongo

Connect to a MongoDB/TokuMX instance instead of a MySQL one. Disables all the MySQL-related captures.

=item --notify-by-email

type: string
Expand Down
Loading