Backup Exec agent not responding

[Trilithon]

Hi,

I'm using a Mac OS X Server ver 10.4.3 and I've installed the Veritas MacOS X agent for remote backup called "be_agnt.tar". We have been able to run succesful backups for some time now (Using Backup Exec 9.1 for Windows Servers installed on a win2k server) but a few days ago I applied a service pack to the backup exec software (SP4) and now I can't connect to the remote agent from my backup exec server. This is only true for backups and restores since I still can make selections of what data to backup thru the agent. So it seems that something is blocking access to the agent service when issuing a backup or restore job. I've re-installed the agent but still no improvement.

Anyone with any ideas or similar problems?

Regards,
T

[sourcehound]

Originally Posted by Trilithon

Hi,

I'm using a Mac OS X Server ver 10.4.3 and I've installed the Veritas MacOS X agent for remote backup called "be_agnt.tar". We have been able to run succesful backups for some time now (Using Backup Exec 9.1 for Windows Servers installed on a win2k server) but a few days ago I applied a service pack to the backup exec software (SP4) and now I can't connect to the remote agent from my backup exec server. This is only true for backups and restores since I still can make selections of what data to backup thru the agent. So it seems that something is blocking access to the agent service when issuing a backup or restore job. I've re-installed the agent but still no improvement.

Anyone with any ideas or similar problems?

Regards,
T

First, make sure the agent proccess is really running (use top)

Second, find out what port it's listening on and see if netstat -an reveals if it's open

Third, you can use LSOF to see if the process is holding an open socket (use the network option) or use can use my utility, Mac HelpMate.

Fourth, you can use TCPflow to monitor the conversation between the agent and the server if one is really taking place or similarly use TCPdump.
I'm going to be in SunnyVale CA presenting some training next week, but I'll have some free time if you want to drop me a line - just download Mac Helpmate and click on the picture and you'll easily be able to reach me - dean [at] macworkshops.com. If we can hook up, I can troubleshoot remotely using a special tool I developed.

[Trilithon]

Originally Posted by sourcehound

First, make sure the agent proccess is really running (use top)

Second, find out what port it's listening on and see if netstat -an reveals if it's open

Third, you can use LSOF to see if the process is holding an open socket (use the network option) or use can use my utility, Mac HelpMate.

Fourth, you can use TCPflow to monitor the conversation between the agent and the server if one is really taking place or similarly use TCPdump.
I'm going to be in SunnyVale CA presenting some training next week, but I'll have some free time if you want to drop me a line - just download Mac Helpmate and click on the picture and you'll easily be able to reach me - dean [at] macworkshops.com. If we can hook up, I can troubleshoot remotely using a special tool I developed.

First, I ran ps -aco pid,command | grep agent.be
which told me that there were 2 pid's running and that is what to expect.

Second, running netstat -a told me that the server is litening on port 8192 which is the right port.

As far as the other suggestions I'll try you helpmate on monday when I get to work. I have a rather limited remote access to the server from home unfortunatly.

One thing that might be worth investigating is the ports used since I recall reading a veritas support article where they mention the agents use of the ndmp port (10000). This port is used my the veritas agent from version 10 and forward. I'm using 9.1 but after upgrading to 9.1 SP4a the troubles began so maybe veritas has incorporated the ndmp port in that update by mistake?

If this port is used by any other third-party software, the agents communication would fail. Now, I know that it isn't used on the Xserve but
I can see that the veritas beremote.exe is utilizing the ndmp port.

I tried to alter the "services" file on both the Xserve and the veritas server so that the ndmp port whas dedicated to port 8192. This did't solve the issue either so I changed it back. Maybe your helpmate can give me some answers...

[Trilithon]

If not, I'll most likely try to contact you next week.
I would greatly appreciate any help.

Thanks
T

[Go3iverson]

I was just about to suggest looking at the ports. The primary concern here is restoration, though. Not to be harsh, but does this solution inspire confidence in you? It doesn't matter how much data you can backup, if you can't restore it, your in some trouble!

I know many more folks who have had this solution fail for them than work in their environments. Of course, you don't want to find out too late, when it becomes an issue of restoration. Also consider the support level you will receive in the case of issue.

As I tell my clients, a great backup solution can keep your job, but the first time it fails to restore, it can cost you your job as well!

[Trilithon]

Originally Posted by Go3iverson

I was just about to suggest looking at the ports. The primary concern here is restoration, though. Not to be harsh, but does this solution inspire confidence in you? It doesn't matter how much data you can backup, if you can't restore it, your in some trouble!

I know many more folks who have had this solution fail for them than work in their environments. Of course, you don't want to find out too late, when it becomes an issue of restoration. Also consider the support level you will receive in the case of issue.

As I tell my clients, a great backup solution can keep your job, but the first time it fails to restore, it can cost you your job as well!

I know that this isn't the best solution and if I 'd been the one to set it up it wouldn't look like this, for sure...

The good news is that I've been asked to evaluate the need for an entirely new backup/archive solution for both environments but until I'm done with that I'm stuck with this so for now, I need to figure something out.

We have a tape library (40 tape AIT robot) that we utilize for both backup and archiving. This rob.lib. is connected to our SAN via FC-AL and managed by the Veritas Backup server. As for now when the agent isn't working, I just copy all the data from one volume on my OS X server to another volume on that server by hand every friday night and this isn't exactly perfect. I need the space on the target volume for other things than temporary backups.

I would like to be able to utilize the rob.lib. for my OS X backups as well, as I did when the agent was functioning.

First, do you know if I can use any built in command to make a full backup of the first volume on friday and on every other night just backup the data that's changed? You mentioned "ditto" last time for copying data but can it do more?

Second, what would you personally suggest I use instead of my current Veritas backup software? I'm leaning towards using diskbased nearline backups and thereafter tape for longterm archiving. I also would like to use the same hardware for both environments (Win & OS X).

Thanks

[Go3iverson]

So, there are lots of options out there. There are built in tools that can handle disk to disk backups for you. Using things like rsync, psync and so forth can help you perform incremental backups.

The issue becomes dumping your data to tape.

There are two ways to consider backup. Some folks want to have a solution that spans all their platforms and needs. Some folks decide on a solution per platform, so that different 'experts' can be in charge of their solution.

Mac OS X has many solutions that provide cross platform backup. The two most popular enterprise solutions are Atempo's Time Navigator and BakBone's NetVault. Both are enterprise class solutions, which include plugins for many popular applications (Exchange, SQL, Oracle, etc). Another solution, which is less costly, but also doesn't quite have the feature list that these other two provide, is Tolis Group's BRU. All three have demos for you to try on their respective sites.

Disk to Disk to Tape solutions are truly the most robust you can implement and are no longer cost prohibitive. Moving your backups to Mac OS X can actually save your company money, as Xserve RAID storage is very low cost and provides high performance.

If you want to just find a way to dump your Mac OS X data to tape in a cheap, efficient manner, consider BRU. If you think that you may want to provide a complete solution for your whole company, Atempo and BakBone will provide things like synthetic/consolidated full backups. The concept behind these is that you take a full backup once, ideally to disk. You then take incremental backups thereafter. Once a week, or on your schedule, you create a new point in time backup, which will combine your last full backup with the incrementals, without having to pull a new full from your live servers. This is exceptionally helpful in completing backups in a short window and also will prolong the life of your tape library.

I may have mentioned this before, but I have a paper that's available on my site (as well as a few others, so don't get confused if you find it elsewhere!). You can get a copy at http://www.district13computing.com/docs/backup3.pdf

Hope this helps!

Michael

[lupobutcher]

Trilithon,
Were you ever able to get this resolved? I am running into the same problem but with BUEexec version 10 and OS 10.4.6. The agent is running, I can select files from the selections dialog in BUExec, but I fail when I attempt to backup.

Any more insight would be greatly appreciated
Lupo