Mercurial > hg > cc > azure

annotate workers/bin/test1.sh @ 13:bba589cab837

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
shrinkJSON.sh: minimise "jq ." output test1.sh: fix lrand regression [_]findPDFs.sh: extract responses with application/pdf Content-Type
author Henry S. Thompson <ht@markup.co.uk>
date Fri, 12 Oct 2018 08:51:50 +0000
parents 36b5d379909a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
1 #!/bin/bash
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
2 # Test script to split CC WAT files across threads
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
3 # to count http: vs. https:
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
4 # Usage: [echo file file_id] | test1.sh id home [-t] numWorkerProcesses
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
5 # If -t, no random wait, just id seconds
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
6 # remove >>errs once tested
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
7 #set -e -o pipefail
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
8 echo $$ > test1.pid
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
9 proc=$1
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
10 res=res$proc
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
11 home=$2
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
12 shift 2
13
bba589cab837 shrinkJSON.sh: minimise "jq ." output
Henry S. Thompson <ht@markup.co.uk>
parents: 11
diff changeset
13 function lrand {
bba589cab837 shrinkJSON.sh: minimise "jq ." output
Henry S. Thompson <ht@markup.co.uk>
parents: 11
diff changeset
14 # cheap bad little random number generator
bba589cab837 shrinkJSON.sh: minimise "jq ." output
Henry S. Thompson <ht@markup.co.uk>
parents: 11
diff changeset
15 echo $(( 1 + ($(openssl rand 1 | od -d | head -1 | tr -s ' ' | cut -f 2 -d ' ') % $1)))
bba589cab837 shrinkJSON.sh: minimise "jq ." output
Henry S. Thompson <ht@markup.co.uk>
parents: 11
diff changeset
16 }
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
17 if [ "$1" = "-t" ]
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
18 then
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
19 shift
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
20 pause=$proc
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
21 else
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
22 pause=$(lrand 60)
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
23 fi
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
24 wp=$1
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
25 touch .running
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
26 function tryRead {
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
27 m=0
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
28 set -o pipefail
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
29 until if [ $((m+=1)) -gt 5 ]; then echo "tried to read $1 5 times w/o success, giving up" 1>&2; return 1; fi && \
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
30 curl -s --insecure -o - "$1"| zcat | fgrep msgtype=response | \
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
31 { egrep -i '"WARC-Target-URI":"https?:' || : ;} > $2
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
32 do
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
33 # try to avoid lockstep retries
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
34 echo ${PIPESTATUS[@]} 1>&2
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
35 sleep $(lrand 10)
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
36 echo $(date) $2 retry number $m 1>&2
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
37 done
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
38 set +o pipefail
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
39 }
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
40 trap "{
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
41 #set -e -o pipefail
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
42 cd $res
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
43 ln -s ../nohup.cc .
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
44 tar -czhf - * | \
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
45 ssh -o StrictHostKeyChecking=no -q $home \"{ cd data/jobs/
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
46 mkdir -p test1.$proc
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
47 cd test1.$proc
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
48 tar -xzf - ; } 2>>errs.1\"
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
49 cd
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
50 rm -rf $res
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
51 ( sleep 5 ; rm nohup.cc ) &
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
52 }" EXIT
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
53 mkdir -p $res
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
54 log=$res/log
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
55 # Don't all start at once
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
56 sleep $pause
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
57 echo \# $(date) > $log
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
58 pRes=0
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
59 while read s id
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
60 do
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
61 url="https://commoncrawl.s3.amazonaws.com/$s"
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
62 export ID=$id
11
36b5d379909a a bit more info in logs
Henry S. Thompson <ht@markup.co.uk>
parents: 10
diff changeset
63 echo \# $(date) "running |$proc|$home|$pause|$wp|$id|" >> $log
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
64 # Experimental retry loop
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
65 tryRead "$url" crawl$id
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
66 if [ -s crawl$id ]
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
67 then
11
36b5d379909a a bit more info in logs
Henry S. Thompson <ht@markup.co.uk>
parents: 10
diff changeset
68 echo \# $(date) $id $(wc -l crawl$id) >> $log
36b5d379909a a bit more info in logs
Henry S. Thompson <ht@markup.co.uk>
parents: 10
diff changeset
69 parallel --round-robin --pipe -j $wp "count1.sh {#} >> $res/{#} 2>>$res/errs{#}" < crawl$id || echo "ppfailed $? ${PIPESTATUS[@]}" 1>&2
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
70 else
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
71 echo "crawl$id empty" 1>&2
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
72 fi
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
73 rm crawl$id
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
74 done < ifile.txt 2>> $res/errs || pRes=$?
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
75 ( cd $res && fgrep -h -v \# [1-9]* ) | tr -d \" > $res/tots
8
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
76 echo \# $(date) main loop exit code=$pRes >> $log
5db6015689a2 slightly updated, 4 W only, versions of old test4.sh and subcount.sh
Henry S. Thompson <ht@markup.co.uk>
parents:
diff changeset
77 rm .running
10
2fbefb8d1a9e wrun.sh: usage catchup
Henry S. Thompson <ht@markup.co.uk>
parents: 8
diff changeset
78