cpd is a pure bash script without any special requirements except a *NIX Box and a not to old bash version.
Furthermore may it still a little GNUish. I can't, and will not, test it with other Unices than ArchLinux but patches and bug reports are welcome.
The purpose is collect copy jobs and start them only if the target drive is not busy by any other copy job under cpd supervision.
There are things which could be improved, or fixed, but I think it is now almost feature complete. So someone would call it a 1.0 RCx release.
cpd [option>...] <command> [<argument>...]
cpd [option>...] newjob <source> <destination>
cpd [option>...] newjob <source>... <dest-dir>
cpd [option>...] newjob -t <dest-dir> <source>...
cpd [option>...] newjob <dest-dir> # Read list of sources from stdin
cpd [option>...] newjob -f <file> <dest-dir>-a Auto start the daemon if not already running
-c <cols> Force output to cut at <cols>, 0 or negative=no cut
-f <file> Read list of source files from <file>
-i <idle-time> Daemon stand by time in seconds, default=600, unlimited=-1
-o <cp-opt> Add <cp-opt> to the called cp command. Disables -r/-m if given
-p <user-prio> Enqueue job with modified default priority. PRIO=6 minus <user-prio>
-s Run a simulation, copy nothing
Only available with GNU copy
-m Merge all files in <dest-dir>. Ignored if -o is given
-r Copy recursive. Ignored if -o is givenh, help [c] Show this help or when c=l License, c=s Source of cpd
H Show more help
newjob <arguments> Enqueue a new job with arguments as shown above
start Start the daemon and job processing
X, exit Terminate the daemon and job processing
P, process Process jobs in foreground or trigger daemon to continue
K, kill <job>... Kill a running or cancel a pending job
p, prio <prio> <job>... Change job PRIO to <prio> or 'PRIO=PRIO minus <prio>' if +-<prio>
r, resume <job>... Resume stopped jobs or re-enqueue old jobs
w, show [c] Show jobs, c=i by ID, c=e errors, c=l job log, c=s daemon status
s, stop <job>... Stop jobs
Y, tidy Tidy up all job dataEnqueue a couple of jobs
cpd newjob /media/a /foo/*
...Assumed you have a symlink cpc->cpd
cpc /media/b /bar/*Process the jobs in the foreground...
cpd P...or start the daemon...
cpd start...and take a look how it is going
watch -n1 cpd wior
watch -n1 cpd w sInstead of calling cp with all job parameter, there will only start a sleep/forced-error job with a crude chosen time of a few seconds.
$ ./test/simulation
Prepare test with clean tmp dir
Run: ./cpd show
ID STATUS PRIO PID SIZE DONE DRIVE TARGET FILES
2 pending 4 - 70M 0% /dev/sdb /media/b/data data1-50;data7;dat~
5 pending 4 - 70M 0% /dev/sdc /media/c/data data2-50;data7;dat~
7 pending 5 - 500 0% /dev/sdd /media/d/notes/ file5-100;file4-10~
1 pending 6 - 512K 0% /dev/sdb /media/b/documents/ paper5-100;paper4-~
3 pending 6 - 32M 0% /dev/sdb /media/b/mediathek song5-5;song4-5;so~
4 pending 6 - 32M 0% /dev/sdc /media/c/documents/ invoice5-5;invoice~
6 pending 6 - 315M 0% /dev/sdc /media/c/mediathek clip5-50;clip4-50;~
Run: ./cpd -s process
Running a simulation, nothing will copied
Start [ 2] [/dev/sdb ] /media/b/data <- data1-50;data7;data6;data5;data4;>
Start [ 5] [/dev/sdc ] /media/c/data <- data2-50;data7;data6;data5;data4;>
Start [ 7] [/dev/sdd ] /media/d/notes/ <- file5-100;file4-100;file3-100;fil>
* New or changed jobs by user *
Stop [ 2] [/dev/sdb ] /media/b/data <- data1-50;data7;data6;data5;data4;>
Start [ 1] [/dev/sdb ] /media/b/documents/ <- paper5-100;paper4-100;paper3-100;>
* New or changed jobs by user *
Done [ 7] [/dev/sdd ] /media/d/notes/ <- file5-100;file4-100;file3-100;fil>
Killed [ 5] [/dev/sdc ] /media/c/data <- data2-50;data7;data6;data5;data4;>
Start [ 4] [/dev/sdc ] /media/c/documents/ <- invoice5-5;invoice4-5;invoice3-5;>
* New or changed jobs by user *
Done [ 1] [/dev/sdb ] /media/b/documents/ <- paper5-100;paper4-100;paper3-100;>
Resume [ 2] [/dev/sdb ] /media/b/data <- data1-50;data7;data6;data5;data4;>
ERROR [ 2] [/dev/sdb ] /media/b/data <- data1-50;data7;data6;data5;data4;>
Start [ 3] [/dev/sdb ] /media/b/mediathek <- song5-5;song4-5;song3-5;song2-5;s>
Done [ 4] [/dev/sdc ] /media/c/documents/ <- invoice5-5;invoice4-5;invoice3-5;>
Start [ 5] [/dev/sdc ] /media/c/data <- data2-50;data7;data6;data5;data4;>
ERROR [ 5] [/dev/sdc ] /media/c/data <- data2-50;data7;data6;data5;data4;>
Start [ 6] [/dev/sdc ] /media/c/mediathek <- clip5-50;clip4-50;clip3-50;clip2->
Done [ 3] [/dev/sdb ] /media/b/mediathek <- song5-5;song4-5;song3-5;song2-5;s>
Done [ 6] [/dev/sdc ] /media/c/mediathek <- clip5-50;clip4-50;clip3-50;clip2->
All done
Run: ./cpd show e
>>>>> Error log of job 2
What ever was wrong
>>>>> Error log of job 5
What ever was wrongNote: If there are errors to read, they are features, but you may find true bugs
Copy cpd somewhere to your $PATH and create a symlink there at your taste, e.g.
ln -s cpd cpcthat's all.
cpd is inspired by Ambrevar who was looking for a tool similar to this.
All contributors of stackexchange.com, bash-hackers.org, dict.cc and many more.
GNU General Public License (GPL), Version 2.0