Network Timeout 41 error

[rdapilmoto]

Getting this 41 error on a backup client and can't seem to resolve this:

10:14:20.543 [21142] <4> bpbkar: INF - setenv BPSTART_TIMEOUT=300
10:14:20.543 [21142] <4> bpbkar: INF - setenv BPEND_TIMEOUT=300
10:14:20.543 [21142] <4> bpbkar: INF - setenv BACKUPID=sys98396_1071504731
10:14:20.543 [21142] <4> bpbkar: INF - setenv UNIXBACKUPTIME=1071504731
10:14:20.544 [21142] <4> bpbkar: INF - setenv BACKUPTIME=Mon Dec 15 10:12:11 2003

10:14:20.544 [21142] <4> bpbkar: INF - BACKUP START
10:14:20.544 [21142] <4> bpbkar: INF - Estimate:-1 -1
10:14:20.545 [21142] <4> bpbkar: INF - Processing /
10:14:20.583 [21142] <4> bpbkar: INF - Excluded /core by exclude_list entry /core
10:45:05.951 [21142] <16> bpbkar: ERR - Cannot write to STDOUT. Errno = 32: Broken pipe
10:45:05.952 [21142] <16> bpbkar: ERR - bpbkar FATAL exit status = 24: socket write failed
10:45:05.952 [21142] <4> bpbkar: INF - EXIT STATUS 24: socket write failed

Any suggestions/ideas what I need to do? I've got an exclude list setup on the client and verified that I have a good link status on my backup network between the server and client....Thanks in advance for any help thrown my way!!

Systems Analyst
HEB Grocery Company
San Antonio, Texas

 

[PGPhantom]

Client - What O/S?
Media server - Same as client or different? If different, what O/S?
Master Server - What O/S?

Error 24 can be different things depending on the O/S. A broken pipe can be caused by DNS, nwtwork throttling etc.

 

[rdapilmoto]

The O/S for the client, media server, and master are Solaris 8.

Systems Analyst
HEB Grocery Company
San Antonio, Texas

 

[PGPhantom]

There are a number of issues relating to Solaris 8 and the solution in every case is to install the lates O/S patches. Has this been done?

 

[rdapilmoto]

Thanks! Will check into this...

Systems Analyst
HEB Grocery Company
San Antonio, Texas

 

[daFranze]

I have got similar problems:

Netbackup & Media Server:
hp Server with Windows 2000 Netbackup 4.5FP3

Client:
Sun Enterprise 450 with Solaris 7 (recommended patches Dec 19 2003); Unix Client Netbackup 4.5FP3
the client used to be a EDM Backup Client, but we are moving all our clients to Veritas Netbackup

When I start backups the bpstart_notify is executed, the bpend_notify, too.
But the backup fails with:
29.12.2003 13:39:31 - Error bpsched(pid=588) backup of client luke exited with status 41 (network connection timed out)

Any ideas?

 

[michalln]

daFranze

Does the bpclntcmd -pn command work (from the client)?
If you don't get an answer from the master server in a few minutes there is a communicatoin problem.

When we get this problem it's usually resolving issues.
Do all you're servers and clients use the same DNS server?
Is the an answer for ping from the client to the media and master server? abd from the media and master server to the client?

 

[daFranze]

bpclntcmd -pn works fine, I get very quick response from the Server; I can communicate from the Netbackup Admin Console with the client (where you may change some client properties)

I guess, I do not have a network/routing/DNS problem; all my clients use the same DNS Servers (we have a redundancy of 3 Servers), my Solaris 9 Clients, which I setup without ever backing it up with EMC's EDM software work fine (unfortunately EDM is not a System V Package but some tarball, I don't know what to remove and how to uninstall)

So I guess it's something within the EDM/Veritas Netbackup Environment which confuses Netbackup...

Regards
-- Franz
Sorry I'm not a native spaeker, I'm from Munich, Germany - &quot;Home of the Whopper&quot;, oh no, &quot;Home of the Oktoberfest&quot; ;-)

 

[peterherrin]

daFranze-

I had a similar problem with a Compaq server (with HP network card) and the solution was to upgrade the NIC drivers on the master server and client to latest versions.

Peter

 

[PGPhantom]

Good point peterherrin - Network card/drivers are quite often the cause of problems, especially after O/S upgrades when drivers can be overwritten.

 

[daFranze]

@peterherrin

Yes, this might be a problem, BUT my Solaris 8 and Solaris 9 clients backup fine, my Solaris 7 clients (latest recommended patchrelease) won't, so I do not realy believe in &quot;bad NICs&quot; right now...

I guess it's something around with EDM (all Solaris 7 Clients used to be EMC's EDM Backup Clients), but I can't proof this right now... :-(

Regards
-- Franz
Sorry I'm not a native spaeker, I'm from Munich, Germany - &quot;Home of the Whopper&quot;, oh no, &quot;Home of the Oktoberfest&quot; ;-)

 

[daFranze]

just to let you know:

it's the NIC!

I can't believe it, the Sun's hme's are autonegotiated to 100fdx, but Netbackup doesn't love this, the hme's MUST be forced (autonegotiation off) to 100fdx.....

in other &quot;words&quot;:

Regards
-- Franz
Sorry I'm not a native spaeker, I'm from Munich, Germany - &quot;Home of the Whopper&quot;, oh no, &quot;Home of the Oktoberfest&quot; ;-)

 

[PGPhantom]

Autonegotiation is the defacto standard under all O/S's that I am aware of - Makes the NIC's more universally acceptable. As a ROT (Rule of Thumb) always change all NIC's to 100 full. The only time you do NOT change is if you are using gigabit as that is automtically Full duplex and you get 1000 instead of 100 - I learnt this one the hard way.

 

[daFranze]

hmm, my experience with Sun NICs ans CISCO network-components is: 100 Megabit Ethernet works fine with autonegotiation on (they seem to have both and the same specification of RFC 802.whatwasit?) BUT our ge-NICs on Sun never get a link with autonegotiation on
btw, ge-NICs do not have 100fdx mode
# ndd -get /dev/ge \? | grep adv_
adv_1000autoneg_cap (read and write)
adv_1000fdx_cap (read and write)
adv_1000hdx_cap (read and write)
adv_pauseTX (read and write)
adv_pauseRX (read and write)
#


Regards
-- Franz
Sorry I'm not a native spaeker, I'm from Munich, Germany - &quot;Home of the Whopper&quot;, oh no, &quot;Home of the Oktoberfest&quot; ;-)

 

[sofiabpm]

From what I see here error 41 is the a network connection failure, but you getting socket errors, which points to me that you have problem with
1. /etc/services file - look for extra spaces in the area of the netbackup entries or missing entries also
2. your machine is not responding to the calls of the master due to is you have 100f.d. on the NIC autonegotiation set and on the switch/router you have 100 f.d without autonegotiation.
3. your machine is talking trough a firewall and some ports are closed
4. make sure bp.conf file is the same as to any other machine which backup successfully and is with the similar software/hardware.
I didn’t read all of the entries but all of the above I see on everyday basis.
Good look

 

[tribalmink]

I recently had this problem on our network of 200 clients (all flavours) - a useful utility to check out the network interfaces is iperf.

It has a valuable -d flag which will highlight any full/half duplex issues with your NetBackup clients - 99% of the time if you eliminate these issues first the error 41's will go away

 

[rdapilmoto]

Thanks tribalmink!
Where can I download this utility from?

Systems Analyst
HEB Grocery Company
San Antonio, Texas

 

[PGPhantom]

Do a search in Google or to save you time

http://dast.nlanr.net/Projects/Iperf/

 

[rdapilmoto]

Thanx, PGPhantom!

Systems Analyst
HEB Grocery Company
San Antonio, Texas