bullnose Posted July 15, 2004 Report Share Posted July 15, 2004 I have been running Retrospect Backup 4.3 very successfully for 4 years (yes I know - the upgrade to OS X and Retrospect 6 is scheduled for November). I currently use a G4 Quicksilver with OS 9.2.2 as the backup computer, running a daily backup of 12 clients (one G4 workgroup server, 9 other Macs and 2 PCs) with a total of 22 volumes. Since last week, the daily backup has been failing on a number of clients with error 519 Network Communication Failed. Any subsequent volumes on the affected client then produce error -1028 Client Is Not Visible On Network. After a backup failure, each affected client machine has disappeared from the network (on all protocols) as if the ethernet cable has been pulled out, and has to be restarted in order to re-establish the network connection. Retrospect continues to process all the clients in the script until the script is complete, and then shuts the backup computer down as normal, i.e. it does not crash. It does not affect the same clients every time - there seems to be no pattern. During the last 8 days since the problems started, only 2 of the clients (1 Mac and 1 PC) have so far escaped the errors completely and have backed up OK every night. The G4 server is the most frequently affected, but has been backed up successfully twice. The backup computer and clients do not appear to have any trouble connecting or filesharing in normal use using either AppleShare or TCP/IP - this problem only occurs through use of Retrospect. As the problem is not confined to 1 or 2 specific clients, I have focused my troubleshooting on the common elements. So far has this has been: 1. Change the backup tapes - no difference. (I don't have another AIT drive or SCSI cable to try but I'm fairly sure it's not the drive or cable or tapes as the errors are not media errors.) 2. Change the Ethernet cable to the backup computer - no difference. 3. Reinstall Retrospect from CD and update - no difference. 4. Install Retrospect on another backup computer which uses a different ethernet connection - no difference. 5. Changed the backup order - no difference. 6. Patched the backup computer and all clients to one network hub, isolated from the second hub on the network and all external network links - no difference. 7. As no.6 but using the other hub - no difference. At the moment I don't know if this is a Retrospect issue (compelling, as it only happens with Retrospect), or a network issue which only affects Retrospect. Can anyone help? Many thanks Richard Link to comment Share on other sites More sharing options...
natew Posted July 30, 2004 Report Share Posted July 30, 2004 Hi My first guess would be operating system corruption on the backup server. Can you boot with base extensions and try the backup again? Can you boot from another disk with a clean OS and try it? The other thing you should look into is a bad network cable or any large power sources next to the cables. I have seen machines act in strange ways when a network cable is not quite up to snuff. Thanks Nate Link to comment Share on other sites More sharing options...
bullnose Posted August 4, 2004 Author Report Share Posted August 4, 2004 Thanks Nate, but as I mentioned I have swapped out all the relevant network cables with no improvement, and there are no large power sources near any of the equipment or cables. I have also tried running backups from a different computer on a different connection, and got the same errors, which in my view eliminates a corrupt operating system as a possible cause. Richard Link to comment Share on other sites More sharing options...
natew Posted August 5, 2004 Report Share Posted August 5, 2004 Hi In your original post you mention you are using hubs. Its possible that random network traffic could overload the hub causing Retrospect to lose access to a client. Faulty printers, network drivers etc. have been known to flood the network from time to time. Can you try using a switch or a crossover cable instead? Thanks Nate Link to comment Share on other sites More sharing options...
bullnose Posted August 5, 2004 Author Report Share Posted August 5, 2004 Thanks Nate, we replaced our hubs with a new switch last week and after a few teething problems the problems do seem to be getting better. Hopefully this will be the solution. Cheers Richard Link to comment Share on other sites More sharing options...
waldorfm Posted August 10, 2004 Report Share Posted August 10, 2004 One common problem is that Network administrators often set their switches (cisco) to fixed full-duplex communication. Most computers, also macs, default to auto-negotiate. If there is a mismatch than the result will be very poor performance and communication errors due to errors on the network hardware communication layer. Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.