Unsolved
This post is more than 5 years old
86 Posts
0
2295
HBAnywhere - Networker - configuration jukebox ADIC
This HBAanywhere I have to design a sort of persistant connection between my networker server and AIC Scalar Jukebox with two tape drives. How to do that?
Unfortunately that is not the main problem, but the fact that I have storage node which is not setup properly: it simplu could not see jukebox and two tape drives. It could see only one tape drive, not the second tape driver neither jukebox (tape library) itself.
pereubu
86 Posts
0
March 16th, 2010 05:00
My question is whether I could solve the problem by simply puting automapping enabled. Persistant binding should be ensured that tape0 will be read by OS and jukebox as tape0 not Tape1. The question is how to do that? I do not have to much knowledge of HBA and its documentation is what I need to understand the system.
bbeckers1
2 Intern
2 Intern
•
137 Posts
1
March 16th, 2010 06:00
it would help if you would be able to state a bit more information. what kind of aidc library are you dealing with? for instance the larger aidc i2000's have so-called ioblades on which you have to lunmap tapedrives to systems that are san zoned to the one of the ioblade's two port. Or is this a simpler jukebox that has tapedrives directly connected to the san?
What are the wwpn's of the tapedrives?
The other lun's on both HBA ports are disks (behind WWPN's starting with 50:06...)? Having disk and tape on the same hba is generally a bad idea unless you have specific hba's that can have quality of service on them to separate disk and tape traffic. However still this is something you might see a lot in smaller environments.
pereubu
86 Posts
0
March 16th, 2010 06:00
That might give you the information you are refering:
compadm>inquire
scsidev@0.0.0:DGC RAID 5 0326|Disk, \\.\PHYSICALDRIVE0
S/N: CK200071900247
WWNN=6006016049AD1C00A2CECBF1713BDC11
PORT=00000001
scsidev@0.1.0:DGC RAID 5 0326|Disk
scsidev@0.2.0:DGC RAID 5 0326|Disk
scsidev@0.3.0:DGC RAID 5 0326|Disk
scsidev@0.4.0:IBM ULTRIUM-TD3 64D0|Tape, \\.\Tape1
S/N: 1210266394
ATNN=IBM ULTRIUM-TD3 12102663
94
WWNN=500308C175E80426
WWPN=500308C175E80427
PORT=00000001
scsidev@4.0.0:DGC RAID 5 0326|Disk
scsidev@4.1.0:DGC RAID 5 0326|Disk
scsidev@4.2.0:DGC RAID 5 0326|Disk
scsidev@4.3.0:DGC RAID 5 0326|Disk
scsidev@4.4.0:IBM ULTRIUM-TD3 64D0|Tape, \\.\Tape0
S/N: 1210266406
ATNN=IBM ULTRIUM-TD3 12102664
06
WWNN=500308C175E80423
WWPN=500308C175E80424
PORT=00000001
scsidev@4.4.1:ADIC Scalar 24 310A|Autochanger (Jukebox), \\.\Changer0
S/N: ADIC_1_C00027529
In fact it is a simpler jukebox that has tapedrives directly connected to the san (I believe that is the case).
"Having disk and tape on the same hba is generally a bad idea unless you have specific hba's that can have quality of service on them to separate disk and tape traffic. However still this is something you might see a lot in smaller environments."
The people who have installed are not very proficient, although they are working for one well known international company. I guess that you are correct. Please tell me what could be other options? There is no specific hba's that can have quality of service on them to separate disk and tape traffic. It is smaller environment but not small because it costs fortune. Yes, there are some problems but I do not know whether they caused by that unfortunate design.
bbeckers1
2 Intern
2 Intern
•
137 Posts
1
March 16th, 2010 08:00
why do you state that the storgenode has to see the jukebox as well? Only one system should control the jukebox, either the backupserver (what I prefer) or the storagenode.
- So what is then the inquire output of the storegnode that does only show one tapedrive? The inquire that you supplied shows 2 tapedrives and the jukebox, so I assume that is on the backupserver (which is a windows system as I can see in the \\.\Tape device files)?
- Is that storagenode zoned correctly to the 2nd drive? What OS is used?
Also I don't really get what the actual problem is?
bbeckers1
2 Intern
2 Intern
•
137 Posts
1
March 16th, 2010 08:00
setting autobinding to enabled is preferred. If I recollect correctly otherwise a system needs to be rebooted after binding it to a specific scsi id. With autobinding enabled, just bind persistently by using the already assigned scsi id. Please read the hbanywhere manual page 132 and further: http://www-dl.emulex.com/support/utilities/hbanyware/40/hbanyware.pdf
what does output of "sjisn 4.4.1" show? That should report how the layout of the physical library is, also showing the serial numbers of the tapedrives.
For example regard the (fictive) output below, where an adic i2000 jukebox has two drives. sjisn is a tool to easily confirm what the layout of an jukebox is. So on OS level a devicefile with S/N 1234567890 (as shown by inquire) should be configured as drive1 within Networker, while 1234567891 should be drive2.
# sudo sjisn 1.2.3 | more
Serial Number data for 1.2.3 (ADIC Scalar i2000 ):
Library:
Serial Number: ADICxxxxxxxxxxx_LLx
SCSI-3 Device Identifiers:
ATNN=ADIC xxxxxxxxxxx_LLx
Drive at element address 256:
SCSI-3 Device Identifiers:
ATNN=IBM ULTRIUM-TD3 1234567890
Drive at element address 257:
SCSI-3 Device Identifiers:
ATNN=IBM ULTRIUM-TD3 1234567891
pereubu
86 Posts
0
March 17th, 2010 02:00
This is what I see on my storage node, which is not configured as the prints screens show. The other sorage nodes do show the same output networker server shows.
C:\Documents and Settings>inquire
scsidev@0.0.0:DGC RAID 5 0326|Disk, \\.\PHYSICALDRIVE0
S/N: CK200071900247
WWNN=6006016049AD1C004A9C5A642135DC11
PORT=00000001
scsidev@0.0.3:DGC RAID 5 0326|Disk
scsidev@0.1.0:DGC RAID 5 0326|Disk
scsidev@0.2.0:DGC RAID 5 0326|Disk
scsidev@0.3.0:DGC RAID 5 0326|Disk
scsidev@0.4.0:IBM ULTRIUM-TD3 64D0|Tape
scsidev@5.0.0:DGC RAID 5 0326|Disk
scsidev@5.0.3:DGC RAID 5 0326|Disk
scsidev@5.1.0:DGC RAID 5 0326|Disk
scsidev@5.2.0:DGC RAID 5 0326|Disk
scsidev@5.3.0:DGC RAID 5 0326|Disk
C:\Documents and Settings\cluster.admin>
AS you could see it can not see the jukebox library and the other tape drive.
pereubu
86 Posts
0
March 18th, 2010 04:00
I had two problems: device reordering which is solved (I hope) by putting persistent binding.
1. Networker Server and Storage Nodes show the same output:
C:\Documents and Settings\Administrator>inquire
scsidev@0.0.0:DGC RAID 5 0326|Disk, \\.\PHYSICALDRIVE0
S/N: CK200071900247
WWNN=6006016049AD1C003A826E441E35DC11
PORT=00000001
scsidev@0.1.0:DGC RAID 5 0326|Disk
scsidev@0.2.0:DGC RAID 5 0326|Disk
scsidev@0.3.0:DGC RAID 5 0326|Disk
scsidev@0.4.0:IBM ULTRIUM-TD3 64D0|Tape, \\.\Tape0
S/N: zzzzzzzzzzzz
ATNN=IBM ULTRIUM-TD3 12102663
94
WWNN=500308C175E80426
WWPN=500308C175E80427
PORT=00000001
scsidev@4.0.0:DGC RAID 5 0326|Disk
scsidev@4.1.0:DGC RAID 5 0326|Disk
scsidev@4.2.0:DGC RAID 5 0326|Disk
scsidev@4.3.0:DGC RAID 5 0326|Disk
scsidev@4.4.0:IBM ULTRIUM-TD3 64D0|Tape, \\.\Tape1
S/N: yyyyyyyyyy
ATNN=IBM ULTRIUM-TD3 12102664
06
WWNN=500308C175E80423
WWPN=500308C175E80424
PORT=00000001
scsidev@4.4.1:ADIC Scalar 24 310A|Autochanger (Jukebox), \\.\Changer0
S/N: ADIC_1_xxxxxxxxxxx
What do you mean by “Is that storagenode zoned correctly to the second drive?”
2. I had to include a path (1 out 5) to the jukebox as it is configured on the rest of its storage nodes. That was it!
In fact I have 2 more questions.
3. On print screen of my HBA 3.1 version I have on each adapter 5 connections. In literature I have always find 4. Am I missing something?
4. You said that having disk and tape on the same hba is generally a bad idea unless you have hba's that have quality of service on them to separate disk and tape trafic. I have two DS 4700M. I am wondering whether I could use some zoning to acheive a form of QoS? The switch is one zone only.
bbeckers1
2 Intern
2 Intern
•
137 Posts
0
March 18th, 2010 07:00
That was just because you stated the drive was not visible on the system. Was correct SAN zoning in place?
2. I don't exactly understand what you mean with that. What did you do? Configure something on the adic library, or arrange SAN zoning to it?
3. what do you mean with 5 connections? The 5 WWPN's you see? It can be way more depending on what you zone to a system on the SAN it is connected to. I've dealt with environments that had more than 10 direct san connected drives zoned to a system or dealing with more than 7 adic i2k ioblades connected to one and the same server. Then you would be way over 4 connections. I see that you have 4 WWPN's of disk storage devices (emc clariion?) and 1 to tapedrives per hba.
4. the DS4700M are smaller4Gb/s switches aren't they? Zoning cannot arrange QOS for you if it is still over the same hba (I know certain brocade hba's can which requires also brocade switches on a certain firmware level). Normally it would be best to separate disk and tape by having dedicated hba's. Two for disk (for path failover) and one (for bandwith purposes and as a safety measure if one hba dies. However in your case it's just 2 lto3's so bandwith on one 4Gb hba would have to be sufficicient, but if it dies you are unable to access your library). When looking at the WWPN's of the screenshot I also suspect that you have a dual-ported hba? So what it that breaks? No storage, no backup. Nothing. 2 dual-ported hba's would be a nice setup, however costs more money...
The switch is one zone only? Excuse me? You're not telling me everything (everything can see everything, would be a nightmare when a system causes scsi resets) is within one big zone? Or do you mean one fabric consisting out of 2 switches. Best practices for disks would be 2 separate fabrics for failover.
bbeckers1
2 Intern
2 Intern
•
137 Posts
0
March 18th, 2010 09:00
I also see that it is a small scalar 24 library. Doesn't that use have integrated so-called SNC to lun map tape drives to multiple systems? That is involved as well with making sure that all systems get the drives they need assigned.I believe they are 2Gb max, which would still be fast enough for the two lto3 drives behind it.
So you use the adic management console software to configure the lunmapping of tape drives to systems?
The library itself is running in LUN mode meaning that you can configure that one (or both) drives present the autochanger to a system as LUN1? SO it is not possible to assign it to a certain system (like you can do on the much bigger aidc i2000)? I don't realyy like the fact that you also see the autochanger on the storagenode. Another system could then cause the library not to respond anymore when it would have hba problems for instance. Only the system performing tapemounts should have access, however the library itself does not seem to have functionality to prevent others systems to see the autochanger if they can see the tapedrive...