cancel
Showing results for 
Search instead for 
Did you mean: 

A few Unix backups that are perplexing me

smwoodcrafts
Level 6
I have am running NBU on Windows 2003 servers running 6.5.2A. There are a couple of Unix backups that are failing with 23 or 24 errors. I don't have access to the Unix servers but tried a couple of things from the Master server. First thing I did was go to host properties and I am able to connect. I then telneted to each box using 13724 and 13782 and the connection. I am not understanding why I am getting these errors.

smwoodcrafts
1 ACCEPTED SOLUTION

Accepted Solutions

smwoodcrafts
Level 6
I had the owner of the units create a bpcd directory and started the job. He sent me the log file and I noticed a "HOST NAME NOT FOUND" error.  I recognized the IP and had the owner add the media server to the hosts table and the jobs are now running fine. I now remember that the engineer pointed the job to another storage unit to put directly to tape and that storage unit is hosted on the media server that was absent in the Hosts table.

Thanks for your help.

smwoodcrafts

View solution in original post

6 REPLIES 6

RameshG
Level 3
Partner Accredited Certified

This could be related to a known issue with the "TCP Chimney Offload" feature.  This problem effects Windows servers that have Broadcom network adapters and have this Offload feature enabled.

 

Microsof Page on this problem:

http://support.microsoft.com/kb/942861

 

Symantec Page on this problem: This article below talks about the client, may be this is applicable to Master as well. 

http://seer.entsupport.symantec.com/docs/296341.htm

 

Try disabling this feature and see if it helps:

Netsh int ip set chimney DISABLED

rjrumfelt
Level 6

and paste it here removing any applicalbe hostnames or IP's?

bptestbpcd -client <client_name> -verbose -debug

smwoodcrafts
Level 6

11:12:29.416 [3088.4856] <2> bptestbpcd: VERBOSE = 0
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.915: host_cache_size: 200
0x000000c8
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.916: cache_time: 3600 0x00
000e10
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.928: host_failed_cache_siz
e: 40 0x00000028
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.929: cache_time: 3600 0x00
000e10
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.915: host_cache_size: 200
0x000000c8
Below is the output requested. Most IPs are private.
This is for one of the probelm servers.

11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.916: cache_time: 3600 0x00
000e10
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.928: host_failed_cache_siz
e: 40 0x00000028
11:12:29.416 [3088.4856] <2> init_cache: vnet_hosts.c.929: cache_time: 3600 0x00
000e10
11:12:29.494 [3088.4856] <2> vnet_vnetd_service_socket: vnet_vnetd.c.2043: VN_RE
QUEST_SERVICE_SOCKET: 6 0x00000006
11:12:29.494 [3088.4856] <2> vnet_vnetd_service_socket: vnet_vnetd.c.2057: servi
ce: bpcd
11:12:29.510 [3088.4856] <2> logconnections: BPCD CONNECT FROM 10.1.133.2.1320 T
O IP ADDRESS.13724
11:12:29.510 [3088.4856] <2> vnet_connect_to_vnetd_extra: vnet_vnetd.c.179: msg:
 VNETD CONNECT FROM 10.1.133.2.1321 TO IP ADDRESS.13724 fd = 1764
11:12:29.572 [3088.4856] <2> vnet_vnetd_connect_forward_socket_begin: vnet_vnetd
.c.532: VN_REQUEST_CONNECT_FORWARD_SOCKET: 10 0x0000000a
11:12:29.572 [3088.4856] <2> vnet_vnetd_connect_forward_socket_begin: vnet_vnetd
.c.549: ipc_string: /tmp/vnet-09817272640349612737000000000-Twailt
1 1 1
10.1.133.2:1320 -> IP ADDRESS:13724
10.1.133.2:1321 -> IP ADDRESS:13724
11:12:29.713 [3088.4856] <2> bpcr_get_peername_rqst: Server peername length = 14

11:12:29.822 [3088.4856] <2> bpcr_get_hostname_rqst: Server hostname length = 8
11:12:29.916 [3088.4856] <2> bpcr_get_clientname_rqst: Server client name length
 = 8
11:12:30.025 [3088.4856] <2> bpcr_get_version_rqst: bpcd version: 06500000
11:12:30.119 [3088.4856] <2> bpcr_get_platform_rqst: Server client platform leng
th = 8
PEER_NAME = ctcbackup00_nb
HOST_NAME = blackfin
CLIENT_NAME = blackfin
VERSION = 0x06500000
PLATFORM = solaris8
11:12:30.119 [3088.4856] <2> vnet_connect_to_vnetd_extra: vnet_vnetd.c.179: msg:
 VNETD CONNECT FROM 10.1.133.2.1337 TO IP ADDRESS.13724 fd = 1748
11:12:30.181 [3088.4856] <2> vnet_vnetd_connect_forward_socket_begin: vnet_vnetd
.c.532: VN_REQUEST_CONNECT_FORWARD_SOCKET: 10 0x0000000a
11:12:30.181 [3088.4856] <2> vnet_vnetd_connect_forward_socket_begin: vnet_vnetd
.c.549: ipc_string: /tmp/vnet-09876272640350227606000000000-tGaGst
10.1.133.2:1337 -> IP ADDRESS:13724
<2>bptestbpcd: EXIT status = 0
11:12:30.197 [3088.4856] <2> bptestbpcd: EXIT status = 0

rjrumfelt
Level 6

pops out from that output.

Have you requested that logging be turned up on the system?  Is this something the system SA can do?  I've seen some of your posts before and I'm sure you're aware of what to do, but just in case, the SA will need to create a bpcd directory under <install_path>/openv/netbackup/logs, and should add the verbose entry to the bp.conf file located under <install_path>/openv/netbackup

Have the SA send you back any logs that are generated from a backup, and post them here.

If you want to sift through large logfiles, the bpcd on the master could have something that complains about not being able to connect to "blackfin"

Also, we just nailed down a case here where we could telnet to a solaris 10 server to vnetd and bpcd no problems from the master, and we could telnet to 13724 locally on the client, however when we ran a backup, no bpcd logs were generated, nor were there any vnetd logs generated after we created a vnetd log directory.  We checked inetd and it was running also, so everything was telling us the backup should work, but it was not.

We enabled vnetd in standalone mode, and the backups are working - this is a workaround for now until the SA can pin down why inetd is acting funny.  Below is the technote that pointed us in that direction - its for older versions of NBU, but it worked.  I dont know if that is your issue or not, we wont know until you can see if bpcd logs are being generated.

If logs are not being generated from backups that initiate from the master server, have the SA run <install_path>/openv/netbackup/bin/bpcd locally on the machine, and it should generate a bpcd log.  This will let you know that something is not communicating between vnetd and inetd.

Hope this helps.

rjrumfelt
Level 6
that might help if you find no bpcd logs beign generated from backups:

 http://support.veritas.com/docs/256685

It talks about bpcd specifically, but the same method can be applied to vnetd - its even more relevant since your client is running Solaris 8.

smwoodcrafts
Level 6
I had the owner of the units create a bpcd directory and started the job. He sent me the log file and I noticed a "HOST NAME NOT FOUND" error.  I recognized the IP and had the owner add the media server to the hosts table and the jobs are now running fine. I now remember that the engineer pointed the job to another storage unit to put directly to tape and that storage unit is hosted on the media server that was absent in the Hosts table.

Thanks for your help.

smwoodcrafts