Unsolved
This post is more than 5 years old
5 Posts
0
967
November 6th, 2006 03:00
Automated clone hanging with 'ready for reading, idle'
I am running Networker 6.0.2 on Tru64 Unix 4.0G with TruCluster 1.6.
(Old I know, but its what Ive got, and I cannot upgrade it to version 7.x.)
I have installed a new HP MSL6030 LTO Ultrium 2 Tape library, with an integrated Fibre Channel router card. The library contains 2 HP Ultrium 460 drives.
(This replaces an older HP MSL5026 DLT tape library.)
This install seems to work fine for backups and restores.
The issue I have had now on three separate occasions is during automatic cloning of the backup tapes:
Networker seems to hang with 'ready for reading, idle' on the source tape and 'ready for writing, idle' on the destination tape.
I have so far had to shutdown and restart the Networker server software to clear the error. (This is not acceptable, as my customer will be phoning me all the time to fix this when I deploy this solution.)
(I havent tried nsrjb -Hv -f /dev/nrmt1h {and nrmt2h} yet, I will try that next as a less drastic fix.)
I have two questions:
1) Is this somthing which has been fixed by a later release/patch of version 6 ?
(I cannot use version 7 as it needs Tru64 5.1B, which in turn needs a re-config of my SAN to support 5.1 clustering, this is too big a change for just the backup product.)
2) Failing a satisfactory answer to 1) Is there a way of avoiding/fixing this issue in an automated way? I.e. When I deploy this tape library at a customer site, I want either the issue not to occur, or to be able to detect and correct it automatically via a shell script or similar.
Any more info on this issue gratefully received.
(Old I know, but its what Ive got, and I cannot upgrade it to version 7.x.)
I have installed a new HP MSL6030 LTO Ultrium 2 Tape library, with an integrated Fibre Channel router card. The library contains 2 HP Ultrium 460 drives.
(This replaces an older HP MSL5026 DLT tape library.)
This install seems to work fine for backups and restores.
The issue I have had now on three separate occasions is during automatic cloning of the backup tapes:
Networker seems to hang with 'ready for reading, idle' on the source tape and 'ready for writing, idle' on the destination tape.
I have so far had to shutdown and restart the Networker server software to clear the error. (This is not acceptable, as my customer will be phoning me all the time to fix this when I deploy this solution.)
(I havent tried nsrjb -Hv -f /dev/nrmt1h {and nrmt2h} yet, I will try that next as a less drastic fix.)
I have two questions:
1) Is this somthing which has been fixed by a later release/patch of version 6 ?
(I cannot use version 7 as it needs Tru64 5.1B, which in turn needs a re-config of my SAN to support 5.1 clustering, this is too big a change for just the backup product.)
2) Failing a satisfactory answer to 1) Is there a way of avoiding/fixing this issue in an automated way? I.e. When I deploy this tape library at a customer site, I want either the issue not to occur, or to be able to detect and correct it automatically via a shell script or similar.
Any more info on this issue gratefully received.
0 events found
No Events found!
ble1
6 Operator
•
14.4K Posts
•
56.2K Points
0
November 6th, 2006 03:00
It is also possible that something is wrong with your setup, but one can't see it from here (running it manually with some verbosity might help).
One approach you could take is simply to explain to customer that what he has is not supported - period. It means it does not matter if it works or not - no one cares.
Ignoring what I said above, my next question would be - is your server running in cluster and did you run cluster script?
ble1
6 Operator
•
14.4K Posts
•
56.2K Points
0
November 6th, 2006 04:00
the 'we will support one version prior to the current
one' policy ?)
Actually current policy is to support a version for 2.5 years since its release. However EOS for 6.x has been announced quite some time ago.
where can I find it?
Officially not. Usually no company will provide link to something they do not support anymore, but I agree it would be nice to have it somewhere even in unsupported area of ftp. You may wish to check older media kits if you have any and if not support might still provide you that version.
mentioned in Solstice backup as a bug for nsrclone
in 6.1.3 on Solaris platform. It is not clear if this
is a fix, or an outstanding bug, I could do with
looking at the 6.x bug fix lists for the Tru64 or
generic Unix version of Networker. Do these still
exist on EMCs website ?
It is going to be much easier if you can provide LGTpa id for that problem to trace if that has been fixed and to what does it apply.
handerson11
5 Posts
0
November 6th, 2006 04:00
I am running the server as cluster aware. I used the TruCluster asemgr cluster config software to make it aware of the new device special file for the media changer arm. (The drives have the same device special files as before, nrmt1h and nrmt2h. So they did not require changing. )
I had to remove the old Jukebox definition from Networker and then run jbconfig to add the new one in. This worked fine.
I have not re-run any Networker cluster install scripts since swapping in the new tape library, but all of the cluster failover test I tried worked fine. The cluster config software can 'see' the tape drives and arm on both systems so I don't think thats an issue.
I will try a series of manual clone runs with verbosity on, and see if I can provoke it to fail. (It is an intermittent fault, the worst kind.)
I may have to try 6.1.4, but I have had trouble in the past with some of the later point releases of 6.x on Tru64 with TruCluster. (I wish I was using Solaris, but there we are.)
I know EMC don't support it anymore (what happened to the 'we will support one version prior to the current one' policy ?)
Do they still have it available for download? If so where can I find it? When it was Legato one of the great strengths of the support website was the ability to download any prior version of the server, client, businesssuite etc. Even if it wasn't supported, this is a very usefull facility, if only for disaster recovery purposes.
Although I can tell my customer that 6.x is no longer supported, we are on contract to give them 5x7.5 support for their system as a whole, and I know they do not have the budget or inclination to go for a major upgrade of this system, much though I would like them to
I have done a google search for this error, and it is mentioned in Solstice backup as a bug for nsrclone in 6.1.3 on Solaris platform. It is not clear if this is a fix, or an outstanding bug, I could do with looking at the 6.x bug fix lists for the Tru64 or generic Unix version of Networker. Do these still exist on EMCs website ?
handerson11
5 Posts
0
November 6th, 2006 07:00
I know its not a hard and fast rule, but many manufacturers still serve older versions of software & firmware on their websites. Obviously there is a limit to how far back you can go, but 6.x is still widely used. Pulling the older software and documentation feels like a 'stick' approach to promoting upgrades rather than a 'carrot'.
The old files do not even seem to be on their FTP site. This is a shame.
It leaves VARs in a tricky spot with older systems they are still trying to support.
It feels like a policy shift by EMC away from the old Legato.
However this is rapidly going off-topic.
My attempts to get the clone to break manually have so far failed. But I am still trying.
ble1
6 Operator
•
14.4K Posts
•
56.2K Points
1
November 6th, 2006 08:00
handerson11
5 Posts
0
November 6th, 2006 09:00
I dont have a media pack 6.1.4 but I will try asking EMC if they can send me a copy of the relevant CD.
Can I ask you where you looked up the bug fix ? Is this info available on the EMC website?