-
Title
Backup dies with Write operation failed 'Service is down' RDA Library error code: 39 -
Description
Backup for one or more clients fail with "Backup dies with Write operation failed 'Service is down' RDA Library error code: 39" -
Cause
RDA Library error code: 39, means there's a communication issue between the Server, Client and DR. There are different reasons for this but usually falls on one of these two branches:
1. Network.
2. RDA Library.
-
Resolution
FIX:
Upgrade to minimum version of netvault 12.2.0.18 where RDA plugin bug QS-1870 fix was applied.
This fix was introduced because of difficulty in manually modifying “TcpMaxDataRetransmissions” on windows servers after windows 2008.
Workaround:
Network:
A . If using Windows, the TcpMaxDataRetransmissions can be increased from the default: 5, to 15 (which is the default on Linux). By creating or modifying the following key on the registry:
HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Tcpip\ParametersValue Name: TcpMaxDataRetransmissionsData Type: REG_DWORD - NumberValid Range: 0 - 0xFFFFFFFFDefault: 5*Before doing any changes on the registry please take a backup or copy from it. Also, please note a reboot will be needed for these changes to be applied.Quest Software does not provide support for problems that arise from improper modification of the registry. The Windows registry contains information critical to your computer and applications. Make sure you back up the registry before modifying it. For more information on the Windows Registry Editor and how to back up and restore it, refer to Microsoft Article ID 256986 “Description of the Microsoft Windows registry” at Microsoft Support.If needed these values could be further increased. On nix the equivalent key would be net.ipv4.tcp_retries2 on /etc/sysctl.conf, once changes are made and saved on the file, they can be applied with "sysctl -e -p".B. If using Jumbo Frames, confirm the MTU is the same across all devices: Server, Client, DR, Switch. Different switches may have different limits. In order to confirm Jumbo Frames a ping can be done without defragmentation and specifying the Jumbo Frame payload. If this fails, something may be wrong with the Jumbo Frames. Examples below assuming a 9000 Jumbo Frame:
For Windows: ping -S <source_address> -f -l 9000 <destination_address>
For Linux: ping -I <source_interface> -M do -s 9000 <destination_address>
RDA Library
As of NV 10.0.5.x the RDA library is included as part of the platform installation. If there's an improper or corrupt NetVault installation, this could be another cause for this error and reinstalling NV would be another possible solution.
The DR OS version, whenever possible should always match the RDS Plugin on the client. An older RDS Plugin may still work with a newer DR OS. A newer RDS Plugin than the DR OS would give some issues.