From ca90b8ddbd4ea771b5d47c97a1bd1ecae9ae60de Mon Sep 17 00:00:00 2001
From: Per Cederqvist <ceder@lysator.liu.se>
Date: Fri, 10 Nov 2006 11:25:19 +0000
Subject: [PATCH] Handle warnings and parallell backups.
Better handling of warnings.
* backup-all: New option: --retry. Only run df on the specified
partitions. Added handling of a "hold" file.
* backup-one-task: New option: --retry. Added -warn, -nowarn and
-neverwarnfree status files.
* lysrdiff-status: Print number of warnings per partition, and list
all warnings.
Support starting two "backup-all" at the same time.
* distribute-tasks
(write_task_lists): New argument: partitions. Only output data for
the specified partitions. Create a lock directory to ensure that
two distribute-tasks processes cannot clobber the same file.
(main): Expect partition number on the command line.
Minor.
* Makefile (install-one-task): New target, extracted from install.
---
ChangeLog | 20 ++++++++++++++++++++
Makefile | 6 ++++--
backup-all | 38 +++++++++++++++++++++++++++++---------
backup-one-task | 26 +++++++++++++++++++++++++-
distribute-tasks | 19 +++++++++++++------
lysrdiff-status | 5 +++++
6 files changed, 96 insertions(+), 18 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 808b1d4..83ad99c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,23 @@
+2006-11-10 Per Cederqvist <ceder@sedesopim.lysator.liu.se>
+
+ Better handling of warnings.
+ * backup-all: New option: --retry. Only run df on the specified
+ partitions. Added handling of a "hold" file.
+ * backup-one-task: New option: --retry. Added -warn, -nowarn and
+ -neverwarnfree status files.
+ * lysrdiff-status: Print number of warnings per partition, and
+ list all warnings.
+
+ Support starting two "backup-all" at the same time.
+ * distribute-tasks (write_task_lists): New argument: partitions.
+ Only output data for the specified partitions. Create a lock
+ directory to ensure that two distribute-tasks processes cannot
+ clobber the same file.
+ (main): Expect partition number on the command line.
+
+ Minor.
+ * Makefile (install-one-task): New target, extracted from install.
+
2006-10-30 Per Cederqvist <ceder@sedesopim.lysator.liu.se>
Progress indication. Lock files. More output. Run ssh in batch
diff --git a/Makefile b/Makefile
index 4d18b16..7a48eb9 100644
--- a/Makefile
+++ b/Makefile
@@ -2,9 +2,11 @@ BINDIR=/nobackup/backup.lysator/bin
all:;
-install:
+install: install-one-task
cp backup-all $(BINDIR)/
- cp backup-one-task $(BINDIR)/
cp distribute-tasks $(BINDIR)/
cp fetch-backup-work $(BINDIR)/
cp lysrdiff-status $(BINDIR)/
+
+install-one-task:
+ cp backup-one-task $(BINDIR)/
diff --git a/backup-all b/backup-all
index 51994f7..7d21088 100755
--- a/backup-all
+++ b/backup-all
@@ -1,10 +1,11 @@
#!/bin/sh
usage () {
- echo $0: usage: $0 '[ --failed ] partno...' >&2
+ echo $0: usage: $0 '[ --failed | --retry ] partno...' >&2
}
failed=
+retry=
while true
do
@@ -12,6 +13,9 @@ do
x--failed)
shift
failed=--failed;;
+ x--retry)
+ shift
+ retry=--retry;;
x--*)
usage
exit 1;;
@@ -27,11 +31,15 @@ then
exit 1
fi
-df -h /lysrdiff/0/perm | sed -n 1p
-for lysrdiffpart in $PARTS
-do
- df -h /lysrdiff/$lysrdiffpart/perm | sed 1d
-done
+rundf () {
+ df -h /lysrdiff/0/perm | sed -n 1p
+ for lysrdiffpart in $PARTS
+ do
+ df -h /lysrdiff/$lysrdiffpart/perm | sed 1d
+ done
+}
+
+rundf
for lysrdiffpart in $PARTS
do
@@ -42,7 +50,7 @@ do
fi
done
-/nobackup/backup.lysator/bin/distribute-tasks
+/nobackup/backup.lysator/bin/distribute-tasks $PARTS
total=0
for lysrdiffpart in $PARTS
@@ -55,17 +63,29 @@ for lysrdiffpart in $PARTS
do
while read category subcategory server serverpath
do
+
if [ -f /nobackup/backup.lysator/etc/stop ]
then
echo Stop file found. Exiting. >&2
exit 1
fi
+
+ if [ -f /nobackup/backup.lysator/etc/hold ]
+ then
+ echo Hold file found. Holding... >&2
+ while [ -f /nobackup/backup.lysator/etc/hold ]
+ do
+ sleep 1
+ done
+ echo Hold file disappeared. Continuing. >&2
+ fi
+
/nobackup/backup.lysator/bin/backup-one-task \
- $failed \
+ $failed $retry \
$lysrdiffpart "$category" "$subcategory" "$server" "$serverpath" \
"$ctr/$total $category $subcategory"
ctr=`expr $ctr + 1`
done < /lysrdiff/$lysrdiffpart/perm/lysrdiff/tasks
done
-df -h /lysrdiff/0/perm /lysrdiff/1/perm
+rundf
diff --git a/backup-one-task b/backup-one-task
index 1925881..d19eba4 100755
--- a/backup-one-task
+++ b/backup-one-task
@@ -5,9 +5,11 @@ usage () {
echo Example: $0 0 home ceder inservitus /export/home/ceder "1/123" >&2
echo Recognized options: >&2
echo ' --failed Only run failed backups' >&2
+ echo ' --retry Only run backups with output from rdiff-backup' >&2
}
failed=0
+retry=0
while [ $# -gt 1 ]
do
@@ -15,6 +17,9 @@ do
x--failed)
shift
failed=1;;
+ x--retry)
+ shift
+ retry=1;;
x--*)
usage
exit 1;;
@@ -64,6 +69,12 @@ then
exit 0
fi
+if [ $retry = 1 ] && [ "`(wc -c < \"$rdifflogfile\") 2>/dev/null`" = 0 ]
+then
+ rmdir "$lockdir"
+ exit 0
+fi
+
if [ "$server" = manhattan ]
then
rmdir "$lockdir"
@@ -133,12 +144,25 @@ then
rm -f "$base"/last-failure
rm -f "$statebase"-fail
- if [ `wc -c < "$rdifflogfile"` != 0 ]
+ if [ `wc -c < "$rdifflogfile"` = 0 ]
then
+ touch "$statebase"-nowarn
+ touch "$base"/last-nowarn-backup
+ rm -f "$statebase"-neverwarnfree
+ rm -f "$statebase"-warn
+ else
echo ${msg}: WARNING: output from rdiff-backup:
echo
sed 's/^/ /' "$rdifflogfile"
echo
+ if [ ! -f "$statebase"-warn ]
+ then
+ touch "$statebase"-warn
+ fi
+ if [ ! -f "$statebase"-nowarn ] && [ ! -f "$statebase"-neverwarnfree ]
+ then
+ touch "$statebase"-neverwarnfree
+ fi
fi
else
mv "$base"/backup-attempt-start "$base"/last-failure
diff --git a/distribute-tasks b/distribute-tasks
index 6b41b48..0b68c3f 100755
--- a/distribute-tasks
+++ b/distribute-tasks
@@ -3,6 +3,7 @@
import os
import sys
import random
+import sets
ROOT = "/lysrdiff"
@@ -98,7 +99,7 @@ def read_new_tasks():
ordered_tasks = new_tasks + ordered_tasks
return new_found
-def write_task_lists():
+def write_task_lists(partitions):
jobs = ordered_tasks[:]
# Pick a few lucky jobs and move them to the front of the queue.
@@ -112,23 +113,29 @@ def write_task_lists():
files = {}
for job in jobs:
- if job.lysrdiffpart() not in files:
- files[job.lysrdiffpart()] = file(
- tasklist_file(job.lysrdiffpart()) + ".new", "w")
- files[job.lysrdiffpart()].write(job.task_desc() + "\n")
+ part = job.lysrdiffpart()
+ if part in partitions:
+ if part not in files:
+ fn = tasklist_file(job.lysrdiffpart())
+ os.mkdir(fn + ".lock")
+ files[part] = file(fn + ".new", "w")
+ files[part].write(job.task_desc() + "\n")
for lysrdiffpart, fp in files.items():
fp.close()
fn = tasklist_file(lysrdiffpart)
os.rename(fn + ".new", fn)
+ os.rmdir(fn + ".lock")
def main():
global fatal
+ partitions = sets.Set([int(x) for x in sys.argv[1:]])
+
for lysrdiffpart in range(2):
read_tasks(lysrdiffpart)
read_new_tasks()
if not fatal:
- write_task_lists()
+ write_task_lists(partitions)
if __name__ == '__main__':
main()
diff --git a/lysrdiff-status b/lysrdiff-status
index 00039be..81257e4 100755
--- a/lysrdiff-status
+++ b/lysrdiff-status
@@ -17,10 +17,15 @@ do
echo -n ' '1day: `find $part/state -name \*-start -ctime 1 -print|wc -l`
echo -n ' 'Stale: `find $part/state -name \*-start -ctime +1 -print|wc -l`
echo -n ' 'Tot: `ls $part/state/*-end|wc -l`
+ echo -n ' 'Warn: `find $part/state -name \*-warn -print|wc -l`
echo ' 'Err: `ls $part/state/*-fail 2>/dev/null|wc -l`
done
echo
+echo Warnings:
+echo
+grep -h -- '-warn$' $statecache-* | sort
+echo
echo Failures:
echo
grep -h -- '-fail$' $statecache-* | sort
--
GitLab