Like others, I have been having problems with random backup failures due to network IO errors. See this thread for another example:
http://bacula.10910.n7.nabble.com/Win32-FD-Write-error-sending-N-bytes-to-Storage-daemon-td35109.html I understand that this is not a bug in Bacula: its going to be a problem somewhere in the stack on either the client or the storage machine. As someone notes in the thread above, because Bacula transports hundreds of gigabytes over a single TCP channel, something that happens only once every hundred GB has a high probability of breaking a backup while being completely unnoticeable anywhere else. Maybe its a dodgy network card, maybe something in the TCP/IP stack on Windows XP. But that doesn't help solve the problem; I can't afford to swap components until I perturb this out of existence (I've tried the usual remedies). Therefore I'd like to suggest that Bacula support a transport mechanism other than TCP. UDT (http://udt.sourceforge.net/) looks ideal. Its got a mature implementation for Linux, Windows, OS X and BSD. The API is close to a drop-in replacement for TCP, so not a lot of new code, and it could be made an option with TCP remaining the default. UDT is designed for high volume data transfer, and is probably more efficient than TCP over a LAN, which is the typical use-case for Bacula. A Google search for "Bacula UDT" turned up nothing, which is surprising. Am I really the first person to suggest this? Thanks to the devs for a great program. Paul. ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://p.sf.net/sfu/appdyn_d2d_mar _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users