On Mon, 2003-03-10 at 20:36, Gene Heskett wrote: > On Mon March 10 2003 14:13, Eric Sproul wrote:
> >My question: If I do a dd of some amount of data (like dumping > > the first 32K of data as I would if I was saving the label), and > > then see (via mt) that the drive's compression is still off, is > > it safe to conclude that the tape was written with compression > > off? > > > I believe so. If the drive has a front panel indicator of > compression use like mine does, I'd use that for the last word. > I don't believe it has an LED for that, but it's hard for me to see them, as the drive is buried in the dark recesses of a rackmount library with only a small viewing window that is mostly blocked by the robot mechanism. I'll have a look in the manual... <"Jeopardy" music> OK, according to the manual, the only density-related indicator present is the one indicating whether the tape is SDLT110/220 (lit) or SDLT160/320 (unlit). No light for the compression setting. > >BTW, the drive has no DIP switches for setting compression-- > > That would definitely be unusual... The lawyers have attacked the > user stuff in the packaging to the point where it may not be > mentioned (I mean "Now why would the *user* need to know that?), > and I'd come a lot closer to being able to believe that as opposed > to its not having any jumpers or dip switches at all to control its > powerup defaults. > This is a Compaq SDLT320 drive. According to the reference guide: "The SDLT drive ships from the factory with data compression enabled for writing. In this mode, data is always compressed when writing to the tape, but the drive is capable of reading both compressed and native tapes. For the drive to write native data, the data compression setting must be changed through the software. To change the setting, consult the backup application software documentation for the data compression enabling and disabling procedure." Believe it or not, it's software-only. From a visual inspection of the library, the part of the drive visible from the rear has no switches. > > it is > >completely controlled by software. Also, my system is Linux, so > > there are no compression-related device names. I am relying > > strictly on mt to manipulate the drive outside of AMANDA, so it's > > important that I be able to trust its output. > > How are you interpreting the mt report so as to define if > compression is on or off? I don't recall seeing that in plain > english in any mt output report since about RH5.2, and it has > changed a bit. Its concise to the point of being obtuse IMO. Well, GNU mt (the one with the "datcompression" option) would report the compression status if you left out the on/off argument. So that's how I was figuring it. However, the mt-st version will set the compression *on* if you throw it "compression" without also saying "on" or "off". I've switched to using tapeinfo from mtx (at Joshua's suggestion), which reports much nicer information: # tapeinfo -f /dev/sg3 Product Type: Tape Drive Vendor ID: 'COMPAQ ' Product ID: 'SDLT320 ' Revision: '2E2E' Attached Changer: No SerialNumber: 'PMC24Y0940 ' MinBlock:4 MaxBlock:16777212 SCSI ID: 1 SCSI LUN: 0 Ready: yes BufferedMode: yes Medium Type: 0x86 Density Code: 0x49 BlockSize: 0 DataCompEnabled: no DataCompCapable: yes DataDeCompEnabled: yes CompType: 0x10 DeCompType: 0x10 Block Position: 202559 "DataCompEnabled: no" is about as plain as you can get. :) Cheers, Eric
