Am 09.04.2018 um 17:46 schrieb Kern Sibbald:
Hi Kern!
> The output from lsscsi looks odd. From what I see, I am not reassured
> that both the tape drives are actually part one at a time and see if
> physically the right tapes are mounted.
I'm afraid, I didn't fully understand what you meant to say here. But I
assume, you are wondering why the devices of the library are so
"distributed".
That I can explain at least partially: The reason is, that the way the
library is connected to the server is somewhat "unique".
The library is an Overland NEO2000 which, unlike most other libraries,
has a dedicated controller card with its own parallel SCSI connectors.
The LTO1 drive also has its own parallel SCSI connectors. The LTO4
drive, on the other hand, has a Fibre Channel connection. In order not
having to use two different host adapters (SCSI & FC), I installed a
FC-to-SCSI bridge in library. (An ATTO FibrBridge 2390C, relabled by
Overland.) On the server side, I'm using a dual channel FC controller
Emulex LPe 11002.
The LTO4 drive and the FibreBridge are connected directly to one of the
FC controller ports respectively.
Or, as a sketch:
---------. .---------------------
Server | | Library
| +-------------.
.-------+ Fibre Channel | |
| |=================================| LTO4 drive |
| FC HA | | |
| |======## +-------------'
'-------+ || |
| || +-------------.
---------' || | |
|| ,---+ LTO1 drive |
|| | | |
|| SCSI | +-------------'
|| | |
|| | +-------------.
|| '---+ |
|| | Controller |
|| .---+ |
|| | +-------------'
|| SCSI | |
|| | +-------------.
|| '---+ |
|| | FibreBridge |
##=========================| |
Fibre Channel +-------------'
|
'---------------------
But I don't have the slightest clue why the kernel splits up the two
channels of the single FC controller into two non-consecutive controller
numbers (0 and 7). But this number asignment is persitent through system
reboots, though...
> [...]
> My experience on LTO-1 and LTO-4 drives is that 512K buffer sizes get
> quite adequate performance so I am a bit skeptical about your need for
> 1MB buffers, but that said, they should be OK.
Well, that's what I found out with some btape "speed" test runs:
https://hirnfasching.de/2018/02/19/geschwindigkeitsmessung-lto-4-laufwerk/
The blog post itself is in German - sorry for that - but the figures
should be understandable anyway.
OK, I have to admit "significantly" is a bit of an exaggeration when
comparing the 1MB values with the 512kB block size values, but 1MB does
result in a larger throughput after all.
Best regards
Sebastian
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
