On Tuesday 30 January 2007 20:24, Darien Hager wrote: > On Jan 30, 2007, at 1:02 AM, Arno Lehmann wrote: > > As far as I know, the heartbeat is not sent during despooling and / or > > attribute despooling. So, if you have that enabled, try turning it off > > to see what happens. > > I haven't enabled spooling--I'm backing up to a plain old file on the > server, and can see the file getting bigger as the job goes on. It > doesn't seem to be tied to the amount of data being written to the > volume as much as the job run time--enabling/disabling gzip > compression or changing the fileset specification has no effect, > unless it makes the job end normally in under 5 minutes. > > The only constant symptom is the time, but I still happens even when > transferring small files to the SD. Just in between one and another > (again no pattern but the job timing) the FD says the SD closed the > connection, and the SD says "broken pipe". Browsing through TCPdump > output on the SD shows everything looks normal (that is, follows the > existing pattern) until the very last packet, where the SD sends an > additional packet with a RESET flag. > > On Jan 30, 2007, at 4:05 AM, Stephen Carr wrote: > > Do you have access to the firewalls? > > No, but I can bug the people who do via a ticketing system. They said > that they changed their *idle* time limits from 5 to 30 minutes > (which I verified with NetCat) but the problem still continues.
TCP/IP standards call for 2 hour idle time limits. If they are adjusting them down to 5 minutes, clearly there are less resourced being held open by disconnected programs, but they may break a lot of programs, and if they are adjusting the timeout limits down so much, they are probably also mucking with keepalive, so it isn't too surprising that things might not work. Another thing to check for is HP printers on the network, which have the nasty habit of using non-registered ports e.g. 9001, 9002, or 9003, which can cause disconnects by Bacula. If you find printers probing/using those ports, either reconfigure the printers or Bacula (preferably the former since the ports are officially registered by IANA to Bacula). > I > haven't modified the connect timeouts--they should be at the default > 30 minutes. > > -- > --Darien A. Hager > [EMAIL PROTECTED] > > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share > your opinions on IT & business topics through brief surveys - and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
