dns timeout?

[Iwan_Tamimi]

I have a client, it is a rhel server v6, at first I didn't put the master server and media server IP and name on /etc/hosts since they are in the DNS, the backup run OK. Then one day the backup got error 23 or 24, I did test to ping to the master/media server using name, it was ok, but the backup re-run still failed. Then I tried to put the master and media servier name and IP on local /etc/hosts, then the backup running fine again?

I still don't know the caused, then maybe someone can answer my questions?

• Why suddenly the backup failed but if I ping the master/media server using name is ok and if I put the master/name ip/name the backup is ok?
• Is it because the ping can tolerate the delay during the name resolving but the netbacakup could not?
• Any netbackup parameter I need to adjust?

Actually this one not happend to the rhel client only, before also happenned on Windows 2008 and HP-UX client, and this error always can be solved by putting the master/media name and ip to local /etc/hosts. The problem is in very near future will use VCS cluster for the master  we cannot hardcoded the IP on the client.

Thank you.

Iwan

[Marianne · Accepted solution]

Remember that ping is one-way test. NBU connection is two-way communication:
Server needs to resolve client hostname to IP address.
Client needs to resolve Server's IP address to valid hostname that appears in SERVER list.

Better way to test will be nslookup:
1) On server: nslookup <client-name>
2) On client: nslookup <server-IP-address>

Also bear in mind that NetBackup has a default host cache lifespan of 60 minues.

Which can be cleared using:

bpclntcmd -clear_host_cache

PS: Please remember that Standalone master can only be converted to clustered master by a Symantec Consulting partner.
Clients will only know Master by its virtual hostname and IP address.

[Iwan_Tamimi]

Thak you for the reply Marianne.

Actually the master and the media is on DNS but the clients (the clients netbackup network) are not. The backup working fine a day before, Then after I put in the /etc/hosts everything is ok.

It is the posiblitiy that the DNS cannot answer the netbackup in time?

Yes for the Standalone master conversion to cluster will be done by Symantec consulting partner.

Regards,

Iwan

[Marianne]

NBU will respond as fast as DNS does. The default timeout is 5 minutes. You will agree that DNS should be able to respond within this time. Client_Connect timeout can be increased, but best to troubleshoot DNS issues.

Maybe ask DNS admins if there was any wrong during this time. There should be some indication in Event Viewer logs.

[revarooo]

I have seen issues where you have multiple nameservers configured in DNS and if one of those is not responding or slow to respond you will get NetBackup timeouts, even if the other nameserver is fine.

[Iwan_Tamimi]

Marianne thanks for the answer.

revaroo, that's what I am experiencing (we do have multple DNS server, Linux and Windows)

I am still doing some investigation, I will posf if I found any clue.

[Iwan_Tamimi]

Hi All,

Actually what Marianne pointed out is correct ( for may case is: "2) On client: nslookup <server-IP-address>").

What I did just to make sure the reverse-dns for the master server has to work (jest like the "2)")

It happened on the RedHat (Primary) it;'s been setup but not on the Windows (Secondary), after we fixed the DNS on the Windows, the client can be backup-ed without putting the master server ip address on /etc/hosts

Thank you for your help.

Regards,

Iwan