roobieroo Posted May 10, 2012 Report Share Posted May 10, 2012 I have a few clients running 9.0.2.102 that seem to randomly go missing from Retrospect. The server is running 10.7.3 and Retrospect 9.0.2. I can test the address and add them using the direct IP address but multicast shows nothing. There is no firewall running on the client but it does run Parallels on occasion. Needless to say with computer that obtain an address via DHCP this is a real problem since they don't get backed up. Any tips or advice on how to get multicast working properly with my clients or do I have to statically assign all clients I want to use with Retrospect? It looks like multicast is broadcasting on the loopback address but not the primary ethernet port. Does this give any clues? Here's the /var/log/retroclient.log contents 1336662409: Client version is 9.0.2.102 1336662412: sopsFileLoad: gotBytes=1792 stateSig=53436667 firstVers=201 curVers=201 1336662412: NetStart: starting interface thread 1336662412: netCheckNewInterfaces: found new address 127.0.0.1:0 1336662412: ConnStartListen: starting thread ConnStartListen for 127.0.0.1:0 1336662417: netCheckNewInterfaces: found new address 192.168.123.103:0 1336662417: Turning client initiated discovery on 1336662417: ipludAddMembership: adding membership for 192.168.123.103 1336662417: iplud: multicast advertising on address 127.0.0.1 1336662427: IPNSRegister(0): registered: "Clerical","91c861ed1cb9d982","0.0.0.0:0" 1336662427: ConnStartListen: starting thread ConnStartListen for 192.168.123.103:0 1336662429: notifyServerList: Client discovery packet sent to 192.168.123.50 1336662430: bindToValidBootPort: gServerPID has been initialized to 54 1336662431: SopsSetCurrentIPAddress: 192.168.123.103 1336662431: --- UpdateClientStateWithMonitorEntered: clientState:10 1336662436: netCheckNewInterfaces: found new address 10.211.55.2:0 1336662436: iplud: bind() failed with error 48 1336662440: bindToValidBootPort: task_set_special_port succeeded 1336662444: IPNSRegister(0): registered: "Clerical","91c861ed1cb9d982","0.0.0.0:0" 1336662444: ConnStartListen: starting thread ConnStartListen for 10.211.55.2:0 1336662444: netCheckNewInterfaces: found new address 10.37.129.2:0 1336662444: iplud: bind() failed with error 48 1336662450: IPNSRegister(0): registered: "Clerical","91c861ed1cb9d982","0.0.0.0:0" 1336662450: ConnStartListen: starting thread ConnStartListen for 10.37.129.2:0 1336663027: notifyServerList: Client discovery packet sent to 192.168.123.50 1336671869: connTCPConnection: non-stream packet too large: 101517849 1336671869: ServicePurge: serviceID 0 not found 1336672027: notifyServerList: Client discovery packet sent to 192.168.123.50 1336672060: ServicePurge: serviceID 0 not found 1336672070: ServicePurge: serviceID 0 not found 1336672411: ServicePurge: serviceID 0 not found 1336672417: NetGetMacAddr: This system's built-in MAC address is 3c:07:54:52:86:91 1336672418: SopsSetOnDemandServer: Setting to 192.168.123.50 1336672419: SopsSetFlags: theServer.parms.clientflags = 8602 1336672627: notifyServerList: Client discovery packet sent to 192.168.123.50 Quote Link to comment Share on other sites More sharing options...
bcarillon Posted May 13, 2012 Report Share Posted May 13, 2012 FWIW... I have never been able to find/add clients via Multicast. I have always had to add directly. 1 Quote Link to comment Share on other sites More sharing options...
roobieroo Posted July 26, 2012 Author Report Share Posted July 26, 2012 FWIW... I have never been able to find/add clients via Multicast. I have always had to add directly. What do you do about laptops that use DHCP so their IP is frequently changing? Quote Link to comment Share on other sites More sharing options...
Lennart_T Posted July 26, 2012 Report Share Posted July 26, 2012 What do you do about laptops that use DHCP so their IP is frequently changing? If the IP is frequently changing then you have a too small range of IP addresses available. If the range allows at least one address for every possible computer then the address would rarely change. Our laptops have kept their address for years and we use DHCP. Quote Link to comment Share on other sites More sharing options...
demani Posted August 2, 2012 Report Share Posted August 2, 2012 If the IP is frequently changing then you have a too small range of IP addresses available. If the range allows at least one address for every possible computer then the address would rarely change. Our laptops have kept their address for years and we use DHCP. That is still relying on habit for that to work. Better to set up Static DHCP assignments (so they always are handed the same IP address), or to get Multicast working properly on the network (that being said, it's hard to tell if Multicast is failing due to a switch/router configuration issue, or if Retrospect is just not working properly). Quote Link to comment Share on other sites More sharing options...
Lennart_T Posted August 3, 2012 Report Share Posted August 3, 2012 it's hard to tell if Multicast is failing due to a switch/router configuration issue, or if Retrospect is just not working properly). It did not work in Retrospect 8.2, even for a small network of two Macs and a network printer. I don't know about Retroepct 9. 1 Quote Link to comment Share on other sites More sharing options...
David W Lee Posted December 14, 2012 Report Share Posted December 14, 2012 According to the team, Retrospect is asking to use UDP with Retrospect's reserved multicast address 224.1.0.38. But the OS is returning the loopback address just as you pointed out. Is there any special network or security setup on the client Mac or on your network? Just taking a stab, not sure if this video about Retrospect advanced network settings would help or not: http://www.youtube.com/watch?feature=player_embedded&v=bjAVvnuq7bM Quote Link to comment Share on other sites More sharing options...
roobieroo Posted January 4, 2013 Author Report Share Posted January 4, 2013 According to the team, Retrospect is asking to use UDP with Retrospect's reserved multicast address 224.1.0.38. But the OS is returning the loopback address just as you pointed out. Is there any special network or security setup on the client Mac or on your network? Just taking a stab, not sure if this video about Retrospect advanced network settings would help or not: http://www.youtube.com/watch?feature=player_embedded&v=bjAVvnuq7bM No changes made with the advanced network settings ever made a difference. It's not clear why Retrospect sometimes would bind to the correct IP address and other times would grab the loopback but I gave up and set all the machines up with static IP addresses. Fortunately they're all desktops in the location. Hopefully this is fixed with the version 10 client since I have some larger installations that will be getting upgraded before long. Quote Link to comment Share on other sites More sharing options...
hrobinson Posted December 31, 2013 Report Share Posted December 31, 2013 Gee, I'm seeing about the same thing with the 6.0.x server. Nice to see how on top of fixing things they are... Makes me wonder why should really upgrade my server from from a G5 or not (just kidding, I know I have to eventually ... before it dies). The log file (/private/var/log/rectroclient.log): 1388424135: Main: Client version is 6.2.2341388424141: bindToValidBootPort: Unable to bind to valid boot port<SNIP>1388424159: bindToValidBootPort: Unable to bind to valid boot port1388424160: netCheckNewInterfaces: found new address 169.254.15.221:01388424160: ipludAddMembership: adding membership for 169.254.15.2211388424160: bindToValidBootPort: Unable to bind to valid boot port<SNIP>1388424180: bindToValidBootPort: Unable to bind to valid boot port1388424181: netCheckNewInterfaces: found new address 192.168.1.7:01388424181: iplud: bind() failed with error 48 48 is EADDRINUSE. I'm guessing, but it looks like they are messing up the parameters to the bind -- giving zero for the port number instead of the Danz port number AND not using SO_REUSEADDR. It probably worked for a while until something else started opening UDP sockets with a port #0 and every interface IP address. Actually, in this case, it looks like it worked when the machine first booted (presumably because it was first), but failed when DHCP finally finished and changed the IP address for the interface. Oh, and as to the 9.x version, they should not be letting the OS pick the IP address for sending multi-casts: It has no way to know what to use, the application MUST bind the IP address for the sending socket. If the OS picks, the only time it is sure to be right is if there is only one interface (and in the cases that matter there are always at least two). The proper way to do this is to get the list of interfaces from the system and bind to every one. There can be more than one IP address for each interface, too -- that makes it "interesting" for sending non-multicast addresses. Quote Link to comment Share on other sites More sharing options...
roobieroo Posted December 31, 2013 Author Report Share Posted December 31, 2013 For what it's worth, I haven't had a problem with this for quite a while now. I guess it could still be happening but it hasn't been something where I've noticed clients not backing up. Quote Link to comment Share on other sites More sharing options...
backy Posted January 8, 2015 Report Share Posted January 8, 2015 (edited) Sorry to dig up an old thread, but I'm having the same problem. Can't add clients unless I know their IP. Which with laptops, tends to change sometimes... Any further advice on this topic? Would love it if this worked. Retrospect Multi Server 9.5.2.103 Windows Server 2012 R2 Thanks for any updates. EDIT: Sorry, just realized this is in the Mac forum. Thanks anyway. Edited January 8, 2015 by backy Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.