OK it's done. Overlord module: http://func.pastebin.com/kA4Xq2ne Minion module: http://func.pastebin.com/dZ7SDMMZ How to use (example): #!/usr/bin/python import func.overlord.client as fc client_handle = fc.Client("*") remotesource='/etc/hosts' localfolder='/tmp/out' print client_handle.local.getfile.get(remotesource,localfolder) I made tests - on RHEL4 - on RHEL5 - with some minions which don't have the file - with some minions unreachable It seams fine. It's slow for big files, but fine for little configuration or inventory files. The slowness has the same cause as for copyfile: xml-rpc is not fitted for copying files. Could you consider testing it and adding it in the upstream git ? Louis 2010/10/27 seth vidal <skvidal@xxxxxxxxxxxxxxxxx>: > On Mon, 2010-10-04 at 11:57 +0200, Louis Coilliot wrote: >> Hello, >> >> it is not so straightforward if there are more than one minion. >> >> For copyfile, the overlord part uses: >> if data: >> Â Â Â Â Â Â Â Â self.parent.run("copyfile", "append", [remotepath, >> xmlrpclib.Binary(data)]) >> >> This applies to all target minions because it is the same file for all. >> >> I can't do the opposite for getfile, because the action differs from >> one minion to the other (the files can have different lengths) >> >> I need to do something like this because I cut the files in chunks to get them: >> for chunknum in range(chunkslen): >> Â Â Â Â Â Â checksum,chunk = Âself.parent.run("getfile", "getchunk", >> [chunknum, source]).items()[0][1] >> >> It is fine when I restrict the target to one minion. >> >> Do you know how I could do the run on one single minion (replacing >> parent.run), so I just have to loop over the minions ? >> >> I don't. > > > > okay - so I finally got a chance to look at this code and I think I see > the problem. > > So - like all func modules when you run from the overlord you normally > get back a dict with keys == minion and value == data. > > Here's what it seems like you want to do: > > modify overlord module getfile.get to include a target_prefix and a > miniondir kwarg. That way if the user only specifies a single minion > then they can put the file they are retrieving into a specific dir. > > If they are doing a bunch of minions then they can have it make a > miniondir. > > also the target prefix is so you could do something like > > getfile.get(source=/etc/passwd, target=/etc/passwd, > target_prefix=/var/minion-backup/, miniondir=True) > > and have it make: > > Â/var/minion-backup/hostname1/etc/passwd > Â/var/minion-backup/hostname2/etc/passwd > etc etc > > Now - with the self.parent.run(getfile.getchunks) it seems to me what > you want to do is be writing ALL of the files at each time. So keep > updating whatever you get from getchunks() until you don't get anymore. > > Instead of working on it like a loop where you loop over the minions - > just keep calling getchunks() until you have nothing left to get. > > Alternatively, instead of using chunkslen to figure out how many times > to loop - let it be handled by the result of getchunks() being ''. > > > Do you see what I mean? > > I think it's definitely doable. > > -sv > > > _______________________________________________ > Func-list mailing list > Func-list@xxxxxxxxxx > https://www.redhat.com/mailman/listinfo/func-list > _______________________________________________ Func-list mailing list Func-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/func-list
