First of all, all the configurations are talking about the library is directly connected to SAN switch, in our case it is not. Does this create a problem?
No. But the fact that there are zones might as you wish to isolate different SAN events from causing bus resets across the SAN.
If not, I'm specifically asking how I can implement DDS on the server and client sides?
Via jbconfig or jbedit commands.
Before that you will need to determine what logical drives (on OS) are belonging to which physical drive. You will need to load into the drave tape and test OS patch on each system to do that and write down this on paper. We already discussed this earlier on forum; look for mt, sjimm and sjirdtag keywords.
Asuuming test1 is Sol. client, can it be something like test1:/dev/rmt/0 ??
No, use 0cbn instead. (if 0cbn maps back to that physical drive).
Remember that order of devices defined by jbconfig has to be order as seen by library (sjirdtag).
And do I need to run jbconfig on the dedicated storage node?
You run jbconfig only from host controlling robotics.
I did DDS configuration using jbconfig, and shared the two drives in C4 autoloader. Now in addition to \\.\Tape0 and Tape1, rd=solarishost:/dev/rmt/0cbn and 1cbn also appears in device list. I then issued a jukebox reset (nsrjb -HEv) and inventory (nsrjb -Iv) form command line but newer devices also got inventoried and really took a very long time!! Does this make sense to you or is there something wrong??
I actually try to do a SAN backup, as you understand. Having done this tapes shared, does the backup of a Solaris host (also a Dedicated Storage Node) guaranteed to be performed through SAN, not through the network?? How can we be sure of that?
I got a belief that the Networker server (Win2003) should somehow 'see' the SAN. Am I correct saying that, because no symptoms that Win2003 sees the SAN luns in the OS level. Remember that there is no zoning in switch, and switch actually acts like a hub.
I then issued a jukebox reset (nsrjb -HEv) and inventory (nsrjb -Iv) form command line but newer devices also got inventoried and really took a very long time!! Does this make sense to you or is there something wrong??
First time inventory might take some time (each time you have some kind of change)... reset also can take some time when called with -E option too.
Having done this tapes shared, does the backup of a Solaris host (also a Dedicated Storage Node) guaranteed to be performed through SAN, not through the network?? How can we be sure of that?
Backup will go to what is defined in storage node field of the client resource. If Solaris host has itself in storage node list then data should go onto itself via SAN (given that storage node name and client name are the same).
I got a belief that the Networker server (Win2003) should somehow 'see' the SAN. Am I correct saying that, because no symptoms that Win2003 sees the SAN luns in the OS level. Remember that there is no zoning in switch, and switch actually acts like a hub.
In "hub" concept your server will see the drive of course, but it should not use it. With DDS NW will now that physical drive phy1 is same as server:logical1 and storage node:logical1. This is because they share same hardware id defined during the configuration. When one drive (logical) will be active second one will appear as busy.
ble1
4 Operator
•
14.4K Posts
0
June 1st, 2006 08:00
No. But the fact that there are zones might as you wish to isolate different SAN events from causing bus resets across the SAN.
Via jbconfig or jbedit commands.
Before that you will need to determine what logical drives (on OS) are belonging to which physical drive. You will need to load into the drave tape and test OS patch on each system to do that and write down this on paper. We already discussed this earlier on forum; look for mt, sjimm and sjirdtag keywords.
No, use 0cbn instead. (if 0cbn maps back to that physical drive).
Remember that order of devices defined by jbconfig has to be order as seen by library (sjirdtag).
You run jbconfig only from host controlling robotics.
atatury
1 Rookie
•
15 Posts
0
June 6th, 2006 01:00
I did DDS configuration using jbconfig, and shared the two drives in C4 autoloader. Now in addition to \\.\Tape0 and Tape1, rd=solarishost:/dev/rmt/0cbn and 1cbn also appears in device list. I then issued a jukebox reset (nsrjb -HEv) and inventory (nsrjb -Iv) form command line but newer devices also got inventoried and really took a very long time!! Does this make sense to you or is there something wrong??
I actually try to do a SAN backup, as you understand. Having done this tapes shared, does the backup of a Solaris host (also a Dedicated Storage Node) guaranteed to be performed through SAN, not through the network?? How can we be sure of that?
I got a belief that the Networker server (Win2003) should somehow 'see' the SAN. Am I correct saying that, because no symptoms that Win2003 sees the SAN luns in the OS level. Remember that there is no zoning in switch, and switch actually acts like a hub.
Regards,
Atatur
ble1
4 Operator
•
14.4K Posts
0
June 11th, 2006 07:00
newer devices also got inventoried and really took a very long time!! Does this make sense
to you or is there something wrong??
First time inventory might take some time (each time you have some kind of change)... reset also can take some time when called with -E option too.
Storage Node) guaranteed to be performed through SAN, not through the network?? How
can we be sure of that?
Backup will go to what is defined in storage node field of the client resource. If Solaris host has itself in storage node list then data should go onto itself via SAN (given that storage node name and client name are the same).
correct saying that, because no symptoms that Win2003 sees the SAN luns in the OS level.
Remember that there is no zoning in switch, and switch actually acts like a hub.
In "hub" concept your server will see the drive of course, but it should not use it. With DDS NW will now that physical drive phy1 is same as server:logical1 and storage node:logical1. This is because they share same hardware id defined during the configuration. When one drive (logical) will be active second one will appear as busy.