James,
The biggest consideration which I know of, and have seen on here many many
times, is the throughput capability of a Linux server. Your main thought
here is not so much how many PCI slots are available but how many pci buses
are available, as this is where your internal I/O limitations fall
I think your easiest solution, seeing as you have some idea of what the
server would need to be capable of is just to talk to IBM and get them to
spec out a suitable solution for you. The reason for this is that there are
so many different solutions to do the job, but I would assume you have a
I have had a lot of bad experiences using imation DLT cartridges in the past
but I have no experience of their LTO products. As for LTO we use TDK's with
a pretty much flawless track record.
Many thanks
Andy Wilcox
UNIX Systems Administrator
Aquila Networks
-Original Message-
From:
I suppose this could depend on whether or not it has actually found any
inconsistencies. Presumably you are running an audit DB because you do have
problems anyway, so stopping it wouldn't be in your best interest.
Anyway I am fortunate enough to have a test environment with a very small
DB. I
Maybe on obvious point but are you ettin useful entries in the AIX error
log?
Many thanks
Andy Wilcox
UNIX Systems Administrator
Aquila Networks
-Original Message-
From: Zoltan Forray/AC/VCU [mailto:[EMAIL PROTECTED]
Sent: 05 March 2004 14:30
To: [EMAIL PROTECTED]
Subject: Re: Can't
Miles,
I think you need to give a little more info as the two scripts sound like
local scripts to yourselves. You will need to give the content of your
scripts to be useful...
eg:
query script q_proc_stats f=d
query script q_ses_stats f=d
... and look at what follows Command:
Many thanks
I know this is only a minor thing but I sure am miffed as to where these
scripts come from.. I don't have them as defined scripts within TSM and
there are no sample scripts in the server directory... I have even trawled
through the media and cannot find them??? Any one have any ideas at what
Hiya,
I think I can see what you are trying to get at, but what I am curious about
is whether the 20MB/s from your disks is due to a physical SCSI bus
restriction, eg your disks are all attached on one 20MB/s SCSI bus, or
whether your 20MB/s is a speed you've seen through some monitoring tool,
It is also worth noting that on every tape unload the LTO drives have a
small mechanical brush that runs over the heads. This is probably the main
reason for the infrequent cleaning requirement.
Many thanks
Andy Wilcox
UNIX Systems Administrator
Aquila Networks
-Original Message-
From:
Always a pig this one but in terms of number of buses the H70 only has one
PCI bus according to the 11th edition of the adapter Placement Guide,
although I have worked with them in the past and I though there were two
buses, with the at least the first two PCI slots on a different bus to the 6
: Wilcox, Andy [mailto:[EMAIL PROTECTED]
Sent: Tuesday, January 27, 2004 16:32
To: [EMAIL PROTECTED]
Subject: Re: AIX adapter placement
Always a pig this one but in terms of number of buses the H70 only has one
PCI bus according to the 11th edition of the adapter Placement Guide,
although I have
We have done exactly what you are suggesting on AIX 4.3.3 ML10. The only
thing we did find beneficial was to set an alternative etherchannel address.
Without doing this we used to get a lot of issues on the network switches
(high end Cisco ones, sorry I can't be more helpful on this one).
With
What you are saying Sean makes perfect sense and I couldn't agree more but
the big issue anyone trying this faces at the moment is the time it will
take to do this operation. If you only have a very small system say 1TB
with not too much in the way of small fies then it would be very feasable.
This isn't exactly a solution but I have the very same problem as yourself
in finding the best way of doing this, and as you have pointed out, there is
no easy way that we have found. It is lookin very much like we will have to
make our TSM service unavailable for a long-weekend to be able to
Very interesting question here... I have in the past done lots of research
on this issue for where I am currently working (although my assumptions are
based on LTO-1 which only has one uplex FC connector).
Your theory about connecting the three drives in a loop is technically sound
except is an
Food for thought (or at least what I have noticed on a 3584) when TSM mounts
a tape, the libvolume still has the home element id of the slot it came out
of. Therefore when TSM checks in tapes it will cannot use the empty slot
created by the mounted tape.
Cheers
Andy Wilcox
Midrange Services
Hiya Ted,
Just out of complete curiosity, why would you want to run such a script from
somewhere other than on the server itself??? We do something pretty much the
same except the scipt runs from AIX on our TSM server...
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
I can't offer any advice on the 5.2.0.0 and STK query but just as a minor
observation, would it not make more sense to got to TSM r 5.2.1.1 instead of
5.2.0.0 as this is a new installation?
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
-Original Message-
From:
This is very much an awkward point to make as I know I am going to
contradict myself but the thing that is possibly more annoying than the
small plug here and there is the messages discussing the plug???
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
-Original
One consideration to make here is that the IBM recommendation for Fibre LTO
drives is no more the 2 per HBA. Although taking into consideration that
this was with LTO1 drives on 1Gb FC HBAs, I would assume you could get away
with up to four LTO1 drives per 2Gb FC HBA, but only 2 LTO2 drives per
are willing to use it.
-- ... although I could be wrong ...
You are, indeed!
Zlatko Krastev
IT Consultant
Wilcox, Andy [EMAIL PROTECTED]
Sent by: ADSM: Dist Stor Manager [EMAIL PROTECTED]
02.09.2003 14:21
Please respond to ADSM: Dist Stor Manager
To: [EMAIL PROTECTED
Hiya Henrik,
I have recently been doing pretty much what you are trying to achieve. The
problem you have is the biggest issue with the move nodedata command. From
reading across the many articles on here, it appears that the only way to
get rid of the old copypool data is to delete the volumes
Steve,
Your assumption here is exactly right. The amount of data being transfer and
stored within TSM has relevance to whether or not compression is
enabled/disabled on the given client machine.
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
-Original Message-
I am in agreement with you there Lloyd. I too have done SDD upgrades but
never had to go through that much hassle. On one of the early versions of
the SDD you used to have to undefine the vpath devices by taking the volume
group offline and then running vp2hd. you then remove the old SDD software,
Hiya Zosimo,
I work in the environment as yourself where we use the same deviceclass for
onsite and copypool tapes. We have talked about this concept before for
using different deviceclasses (working against different tape technologies).
The only advantage we could come up with is that it
My interpretation is that any server running TSM client software or TSM
server software has to be licensed per CPU. The bonus here is that for those
of us with old tape libraries and a spare server can setup up a legitimately
licensed test TSM environment so long as the spare server is CPU
Not quite the idea you are considering but what we have implemented on AIX
is to have 2 * gigbit ethernet interfaces configured as one device using an
etherchannel config (supported by AIX at 4.3.3 but switch needs to support
capability). One of the advantages is that if either card goes down then
Your question sort of answers itself in that you could create the database
the same size or bigger but you need to have an idea of how big it is in the
first place.
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
-Original Message-
From: Gerald Wichmann [SMTP:[EMAIL
I don't know enough about your configuration for too much comment but in our
scenario which is a 3584, using the Atape driver on AIX I would query libv
within TSM to see what logical labels TSM thinks are in which slots and then
compare this against a listing from tapeutil -f /dev/smc0 inventory,
I found that you can't do a move nodedata against a copypool. You have to
move the data from the onsite pool to another onsite pool and copy this to a
different copypool. The data in the old copypool is then inconveniently left
around (although existing copy group expiration value will apply).
To keep it simple
1. has to be a no, unless you have an alternative method of knowing what is
on your tape volumes... with the volume history file you wouldn't know which
is your dbbackup tape(s)
2. You do not need to know the admin password as you do the restore using
dsmserv interactively
What version of Tru64 are you running?
Cheers
Andy Wilcox
Midrange Services
Aquila Networks Services Ltd
-Original Message-
From: Conko, Steven [SMTP:[EMAIL PROTECTED]
Sent: 26 March 2003 15:57
To: [EMAIL PROTECTED]
Subject: Re: stupid backup compatibility question
okay,
I am not familiar with TSM on Solaris but I have seen a similar issue on
AIX. The issue we had was that we had a 100Mb ethernet adapter set to
auto_negotiate and the Switch (A high-end Cisco piece of kit) was also set
to auto-negotiate. When the network came up, the AIX server set itself to
100Mb
You have had some sessions attempting to connect to your TSM service whilst
the restore was taking place.
When ever I have run dsmserv interactively for maintenance, I get automated
sessions on various clients attempting to connect and I get exactly the same
errors.
If you know of any such
Whether this helps or not, on testing the backups of Oracle databases to TSM
we found it far quicker to set a max filesize (1GB is what we use) on the
disk pool and let the big DB files got straight to tape. Your bottleneck is
then with the setup of your tape drives. We don't backup a given DB of
Firstly I must ask people to accept my apologisies if this message doesn't
look right This is first time I have tried to post a message hear :-)
Just reading through a scenario posted by Michael Wheelock about slow tape
to tape perforance... There are lots of considerations here to take note
36 matches
Mail list logo
