Unsolved
This post is more than 5 years old
79 Posts
0
7653
Bare metal restore wizard fails with 0x80004005
Hi,
I've read a few other threads about this, but I'm thinking I'm facing another problem. I'm trying out the BMR wizard on a physical server. I have a complete level full backup including the DISASTER_RECOVERY save set, I should be all set.
The wizards starts ok and can get as far as browsing available backups for my test server. Disks are listed fine, and it seems to correspond what a critical restore should need.
Before getting to that stage I have deleted the NSR peer information, since I've found it interferes with a restore.
The restore starts, but after a few seconds I get "The restore failed with error 0x80004005. Please see the ossr_director.raw file for details".
I've been looking through that file, and I can see a command:
recover -iY -s networkerserver.domain.local -S 016c1f33-00000006-5b4fc2ec-544fc2ec-2eab1700-7c0877a4 -d ..\temp -q
and then an error "Can't parse OSSR_Info.xml" or something similar. From the recover command, I would expect the OSSR_Info,xml to be restored into nsr\temp, but it's not. No wonder the next step can't parse the file.
If I try the save recover command on a storage node it restores OSSR_Info.xml and OSSR_BCDoc.xml, fine. so I'm fairly sure the save set is valid and browsable.
I have a lingering feeling it has something to do with the physical server having two NICs. Normally, they would be teamed, but that's not possible when using the BMR wizard since it only contains the bare minimum of drivers. Like the backup server is sending the restore to the wrong NIC or something in the authentication chain gets confused.
Also, when looking in the daemon log on the NetWorker server I can see the DISASTER RECOVERY save set being restored successfully, but obviously it didn't make it to the client I'm trying to restore.
I have tried the same wizard on a virtual machine I wanted to try to restore as if it were a physical, and it worked fine. I had load other drivers manually for the NIC and SCSI controller, but the restore was smooth. The only real difference I can think of is that the VM only has one NIC.
ble1
14.3K Posts
0
October 30th, 2014 07:00
The only time I had this error was when I was trying to restore machine as something else and not using original settings. As for NIC, use just one (the one you need for restore) and that will be fine. To verify that it works, you can from CLI check if ping is ok or not.
sechm
79 Posts
0
October 30th, 2014 07:00
Which is curious. I can from the BMR session where I have a minimal cmd window ping the networker server. However, I don't get any ping reply when pinging the recovery client from the networker server. The name resolution is correct, I have double checked. It might be that icmp is not active in the Windows PE environment.
As for only using one NIC, I'm not quite sure how to go about it. It is a blade server, there are no physical NIC ports I can pull out a cable from and unchecking the network port in the blade admin GUI doesn't do anything.
sechm
79 Posts
0
October 30th, 2014 08:00
I have a "Recover of rsid 15922 failed. RPC receive operation failed; errno = Unknown error" in the recover.log file
and now on my second revision of the OSSR_director.raw log "Could not load file X:\Program Files\EMC NetWorker\nsr\tmp\OSSR_Info.xml".
No wonder the parse error is the next error.
It all points to something in NetWorker on my recovery client not being able to properly receive the restored files.
ble1
14.3K Posts
0
October 30th, 2014 12:00
I had the same with ping... one of the musts, at least for me, was to disable firewall in WinPE. I believe docs have that step outlined. As for NIC, fact that you can ping says you are already good with that.
sechm
79 Posts
0
November 12th, 2014 04:00
Answering my own question, but it might help others in the same situation. I just could not let go of my idea that having two NICs somehow got the restore confused Ie requesting the restore from one NIC but transferred to another.
I tried numerous things:
Disabling one switch port in the blade enclosure. Did nothing, both links remained up. I'll have to talk to the server and network teams.
Tried loading newer drivers manually during the Windows PE setup in a hope it would load a teaming driver. Did nothing. Windows PE might not even support teaming.
Tried turning off DHCP. The BMR restore wizard refused to move forward.
Tried disabling one NIC with "netsh interface set interface "Local Area Connection 2" disabled. Did not work. Google suggests it has to do with the command shell not having enough privileges. Really Windows? Tricks to elevate did not apply in Windows PE.
Finally. setting one NIC to the original server's IP configuration and the other to a dummy IP which would not route to the NetWorker server.
Bingo! Now the BMR could start.