Hi, On 2/24/2007 11:43 AM, Pierre Bernhardt wrote: > Arno Lehmann schrieb: > >>>Here my feature request for a cleaning enhacement: >> >>If you really want that to be a feature request you should rewrite it >>into the format suggested by Kern. You can find that on the project web >>site, I believe. >> >> >>>1. It should be possible to declare for one or more cleaning slots >>> in a library. >> >>That would be done in mtx-changer, possibly configurable. This would be >>possible rather easily, I think. > > Hmm... I've read and understood I can change mtx-changer skript by simulating > barcodes. But the "barcodes" will then be use by bacula for label naming for > all tapes I will insert in future, too. > If I have more than 8 cartriges (I have near than 50, ok not all used > for bacula at the moment...) this could be a problem. So what I need > is here 7 Slots should didn't have a barcode label and the 8 will > get a barcode named CLN... > > Could be this a way?
Yup, that would be one way. Personally, I'd rather simply have mtx-changer report one slot less than actually exists, and stick a big label to the device "Slot 7 Cleaning Tape ONLY" ;-) > > >>>2. A max. cleaning procedures for a cleaning tape should declared. >>>3. The cleaning procedure should only arrange by bacula (count the >>> cleaning procedures) >> >>This is encompassed by the existing feature request, IMO. >> >> >>>4. The cleaning procedure should be declared by a job. >> >>No. You really don't want to clean a drive unless the drive itself asks >>you to. > > Oh you don't really understand my intervention. I don't need really a > schedule. But handling cleaning in bacula (not only by bacula) by running > a job by run command will be a nice feature. Ok, I now understand that. > So here we have a good > way to count the cleaning jobs. When you manage cleaning by Bacula, there's no need for that job fpr counting purposes, because the catalog already has a data field for each volume for the number of mounts. For cleaning tapes, that maps nicely to the number of uses... > Repeat: No schedule of cleaning. Possible > checking for cleaning and if neccessary then clenaning if schedule. > > >>>5. The job should be run by a request (tape alert, run command,...) > > I mean there is a way that the drive itself could request for cleaning. > Possible a command which works by that...But this for future idea. Yes, tape alerts in SCSI terminology. Almost all tape drives still running today should support that. Getting the information into Bacula is rather simple (and, in fact, done today with the default config), but doing anything useful with it is left to the user now. >>>6. The cleaning job should be abort if the number of cleaning requests >>> are arrived (message or other) >> >>I don't think that we need extra job types for drive cleaning. I rather >>think that an external script would be best - think of mtx-changer for >>tape movement - or, as a first step, a simple operator intervention request. > > Every cleaning cartrige should only used a maximum count. This is my idea > behind this point. The abort should only then notifies me that I need to > insert a new cartrige. Ok... we seem to have terminology problems only, I suppose :-) I'd simply do the counting by Bacula itself, and introduce no a new job type, but rather a new command - 'clean drive=...' or something. You could run that command manuall, but more useful would it be to have it started automatically when necessary. Then, when Bacula has a cleaning cartridge available, i.e. it knows about one being in the autochanger, it unmounts that device and calls a special cleaning script, or (probably a better idea) a function in mtx-changer which does the actual cleaning. The cleaning process could do: - unload tape from drive if necessary. - load cartridge from slot indicated on command line - watch mtx output and try to find a "cleaning successful" message - check for tape alerts to see if more cleaning attempts seem appropriate or another problem exists. - if necessary, unload cleaning tape to the slot where it came from - return with useful data to Bacula: Return code to indicate success or failure One line of numerical text to report the number of cleaning done with the tape, for Baculas book keeping Text presented to to the user and, in case of an error, mailed to the operator. Bacula itself would have to increment the cleaning tape use counter and mount the drive again. When no cleaning tape is available, Bacula would (on non-critical drive conditions) inform the operator and continue, probably checking for tape alerts more often now. When a critical tape alert is found, Bacula would stop using that drive until the condition is fixed. Along with the 'clean' command, we'd need a command to manually increment the use counter for the cleaning tape. Could even be something like 'clean drive=... volume=...' when used interactively. >>>7. update slots scan should never scan the defined cleaning slots. >> >>Item 1. > > My answer :-) > > > >>>8. cleaning slots should never requests for reading/writing. >> >>Item 1. >> >> >>>What did the tape if I do cleaning manually: >>>(Cleaning cartrige in slot 8) >>>1. mtx load 8 >>>The changer loads the cartrige in the drive >>>The drive inject the cartrige and cleaning process is startet. >>>The drive ejects the cartrige and know its different: >>>Libra-8: mtx unload 8 is needed because the drive eject don't >>> unload the cartrige back in slot. >>>HP 12000: The drive eject the cartrige and the changer unloads >>> the cartrige automatically back in the slot (6) >>> (I'm not reallly shure. didn't use the library any more) >> >>First, drive cleaning is not a daily operation, so the actual loading of >>the cleaning tape could be done manually, as a first step. > > Yes I know. > >>The more important step towards integrated drive cleaning is to >>interface that into the normal tape operation. We will need job pausing >>for this. Just imagine that you have a really huge job running, and >>after the first few blocks written to tape the drive wants to be >>cleaned. Today, you'd have to wait until that tape is no longer busy, >>i.e. until all jobs to that tape are finished, or until a tape change is >>required. Being able to pause the job, unmount the tape, clean the >>drive, mount the tape again and continue the job would be a major >>improvement. > > A cleaning handling in bacula has the nice impact that I can run a job > manually. I can put this in the queue so the running job will be finished, > the cleaning job will be start, then next job will be run. Well, what you say is true, but I don't think it's the best solution. Think of my previous example of a a job that's going to fill more than one large capacity tape. After the first few blocks, the drive requests to be cleaned. I'd really prefer to have the cleaning done as soon as possible, and not only after all other jobs are finished, or manual intervention is necessary anyway. > Again I aggree with you: No cleaning procedure if cleaning is not needed. > But if cleaning needed, than in bacula with a job and a dedicated/configured > slot for that is very good. I still don't see the need for a job, but apart from that I completely agree. > MfG... > Pierre Bernhardt > > Sorry for my bad english :-) I'm not a native speaker, but I don't think it's that bad... Arno > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys-and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users -- IT-Service Lehmann [EMAIL PROTECTED] Arno Lehmann http://www.its-lehmann.de ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
