Student Computing Services

OK, so here's a problem. I've just run an incremental for the vol1: volume. It backed up 204 objects for a total of 303.5 KB taking 1:23. I used the following command:

i vol1:home/

This is the command line command to tell TSM that I want an incremental backup of the home directory on vol1: This worked fine.

Trying to continue my testing I tried to restore vol1:/home to the copy: volume, so I issued the command:

res vol1:home/ copy:tsm/

This tells TSM to restore everything it has for vol1:home/ and put it on the copy: volume in a directory called TSM. This didn't work. I got the following error:

"ANS1084 No files have previously been backed up for vol1:home/"

Thinking I may have had a syntax error, I tried restoring sys:system to the copy: volume. I issued:

res sys:system/ copy:/ This worked just fine.

I then tried various other ways to restore the previous vol1: information, but always got the same error.

I have started the restore via the web client although I've been told by TSM support that this isn't the best way to do restores.

The TSM incremental finished in 10 minutes. It backed up 78.3 megs.

For reference the last incremental for this volume finished 20:09:02 10/25/05.

There are too many variables to determine with any certainty the huge discrepancy between Veritas and TSM. The test will be repeated this afternoon. Now that I have a baseline, I should be able to get some better reading for everyone.

I have been testing a trial version of veritas. I did a full backup of the live locker data. this morning I did an incremental. This took 37:13 and backed up 4.3 gigs. I am doing the same on the TSM side right now and will post the results later.

For reference, the Full for this volume ended at00:02:42 on the 10/27/05

So the backup job that started yesterday has been stopped. TSM has again hogged the cache memory causing me not only to stop the job, but restart the system.

I have deleted the home/home dirs that were on the copy and restore volumes.

I have started an incremental of the copy volume.

Here is the Restore from the Copy volume to the Restore volume. It completed

Objects restored 951,198
Objects Failed 0
71 gigs
2 hrs 2 mins. Data Transfer Time: 4,424.29 sec
Netware Data Transfer Rate: 16,856.59 KB/sec
Aggregate data Transder Rate: 11,261.67 KB/sec
Elapsed Processing Time: 2:02:40

Started the restore of copy to restore via the command line. The last two from the web client didn't start.

Restore process of the copy volume to the restore volume has started.

Another restore has been initiated. I checked to see which tsa nlm's were loaded and the were TSANDS and TSAFS. The latter is the one that is supposed to be used now instead of TSA600.

The three day restore went down. I feel this is because of a cache memory allocator problem. Kevin has been having problems with this when he's been trying restores for CNS.

I just stopped the second dsmc thread that was running - trying to get the backup to run a little faster.

The job is still running. Currently it has backed up about 500 MEGS. This is day 3 of the backup job.

On a side note. I downloaded Veritas and did a backup to disk in about 57 minutes. I realize this is going to disk, but I will test backing up to DLT tape shortly.

Stopped the restore process that was started yesterday. I have just started another restore via the dsmc command line.

The error log had the following at the bottom:

10/19/2005 16:08:44 Error writing to http socket.
10/19/2005 16:08:44 Error -50 sending ht request
10/19/2005 16:08:44 Error writing to http socket.
10/19/2005 16:08:44 Error -50 sending ht request
10/19/2005 16:08:44 Error writing to http socket.
10/19/2005 16:08:44 Error -50 sending ht request
10/19/2005 16:08:44 Error writing to http socket.
10/19/2005 16:08:44 Error -50 sending ht request
10/19/2005 16:08:44 Error writing to http socket.
10/19/2005 16:12:06 Error -50 sending ht request
10/19/2005 16:12:06 Error writing to http socket.
10/19/2005 19:17:54 ANS1005E TCP/IP read error on socket = 16, errno = 54, reason : 'Connection reset by peer'.
10/19/2005 19:17:55 sessSendVerb: Error sending Verb, rc: -58
10/19/2005 19:17:55 sessSendVerb: Error sending Verb, rc: -58
10/19/2005 19:17:55 Operation stopped by user
10/19/2005 19:17:55 sessSendVerb: Error sending Verb, rc: -58


The "connection reset by peer" is a bit odd. I didn't stop it, so I'm not sure what it's talking about here. Either way, I've started the backup again. I did start two threads from the dsmc command line, but I'm not sure if this will actually give me two threads or not.

Updated the client to 5.32

The backup has stopped. I got a TCP error. I have tried a couple of time now to restart the job, but I can't get things going. I click on the plus side next to LOCAL, but nothing happens. A piece of the error log follows:

10/19/2005 09:27:51 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:27:52 sessSendVerb: Error sending Verb, rc: -50

I have unload the dsmcad.nlm and reloaded it to maybe help things along. Again I am using tsafs.nlm and tsands.nlm modules.

I have tried clicking the + sign next to the local heading. I'm still waiting. I also unchecked the "Use less Memory" option.

During this period the error log shows the following:

10/19/2005 09:41:38 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:39 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:40 Error -50 sending ht request
10/19/2005 09:41:40 Error writing to http socket.
10/19/2005 09:41:40 Error -50 sending ht request
10/19/2005 09:41:40 Error writing to http socket.
10/19/2005 09:41:40 Error -50 sending ht request
10/19/2005 09:41:40 Error writing to http socket.
10/19/2005 09:41:40 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:41 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:42 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:43 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:44 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:45 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:46 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:47 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:48 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:49 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:50 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:51 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:52 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:53 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:54 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:55 sessSendVerb: Error sending Verb, rc: -50
10/19/2005 09:41:56 sessSendVerb: Error sending Verb, rc: -50

I have loaded the tsands.nlm to see if this helps.

Stopped the backup of the copy volume. After 20+ hours TSM only backed up 28 gigs.

I am restarting the backup process via the web client to see if I get any better speeds.

John Wang just called to review the problems. He wants to hear after we get this next test finished. The test is to backup the "copy" volume and the restore to the "restore" volume.

DSMerror and DSMsched logs have been reset. This is to clear out the logs for the next restore process.

Andra xcopied everything from the true Locker space to the copy volume on Locker. I have just started a backup of the data on the volume. I did this via the command line on Locker.

Another error. I don't know what this one was though, it went away to quickly.

Forth Restore Started. This time I am replacing all the files.

Web interface locked up.

Third restore process started.

Disk full again after 5661 files. User was nssmith2.

Quotas removed. I've started the second restore job.

I am currently removing the quota that was on the student's container. This should stop any more Disk Full errors.

It seems as though once TSM sees an error like this, even though it's on one directory, the program ceases to restore anymore files.

Started the TSM testing prcess.

1. Restarted server to clear two abends and let the server find the new storage.
2. Created the swimming storage pool
3. Created two volumes - Copy and Restore
4. The first step is to copy everything from the current home directoy from this date to the Copy volume on Locker. This will give a good working copy to test with.

I started a restore process to the Restore volume on Locker. The Point in time is set to Sept 19 at 14:00. I got a "Disk Full" Error on the first attempt after 5661 files inspected 5660 restored and 1 file error and 980.91 MB restored.