-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Performance of first synthetic full to tape?
[Edit: Added that this copy has 28 restore points]
Doing all sorts of testing heading towards my "grand plan" of bullet-proof backups (backups & copies as well as tape copies across 2 sites).
I have a backup copy (2tb full, 28 restore points) to our second site and wanted to test sending that to tape at that site as well - If I can get tape working as I want to at both ends then we can get rid of our Iron Mountain contract that costs us thousands of £/year.
As it's a first run it's creating a Synthetic Full but it seems to be taking far longer than I expected even to just get started. (I'd have thought that as it's a backup copy/reverse incremental it would have just dumped the most recent .VBK file to tape to start with?).
At this point it has been running for 1.5hrs and generated a .VSB file of 4.17Gb (so far) but nothing has started writing to tape as yet but, more importantly, there's absolutely no indication whatsoever of progress, e.g. I don't know if it's going through each .VIB in sequence and, if so, how far in it has got so far (I might have a dig around the current log file to see if there's any info in there).
So, I know things can vary greatly based on hardware but is there a ball-park timescale I can consider?
Using V8 Patch 2 (3 going in this week, v9 on the cards). Tape server is physical, this particular repository is 4 x 4TB SAS in RAID-5 (potentially moving to a 12-disk FC array).
Once it actually gets going I expect performance to be pretty decent, we can max out our LTO-6 drive reading from these disks but I do appreciate that there are lots of IOPS associated with a synth full, it'd just be VERY useful to have a vague clue as to real-world progress...
Paul
Doing all sorts of testing heading towards my "grand plan" of bullet-proof backups (backups & copies as well as tape copies across 2 sites).
I have a backup copy (2tb full, 28 restore points) to our second site and wanted to test sending that to tape at that site as well - If I can get tape working as I want to at both ends then we can get rid of our Iron Mountain contract that costs us thousands of £/year.
As it's a first run it's creating a Synthetic Full but it seems to be taking far longer than I expected even to just get started. (I'd have thought that as it's a backup copy/reverse incremental it would have just dumped the most recent .VBK file to tape to start with?).
At this point it has been running for 1.5hrs and generated a .VSB file of 4.17Gb (so far) but nothing has started writing to tape as yet but, more importantly, there's absolutely no indication whatsoever of progress, e.g. I don't know if it's going through each .VIB in sequence and, if so, how far in it has got so far (I might have a dig around the current log file to see if there's any info in there).
So, I know things can vary greatly based on hardware but is there a ball-park timescale I can consider?
Using V8 Patch 2 (3 going in this week, v9 on the cards). Tape server is physical, this particular repository is 4 x 4TB SAS in RAID-5 (potentially moving to a 12-disk FC array).
Once it actually gets going I expect performance to be pretty decent, we can max out our LTO-6 drive reading from these disks but I do appreciate that there are lots of IOPS associated with a synth full, it'd just be VERY useful to have a vague clue as to real-world progress...
Paul
-
- Product Manager
- Posts: 14716
- Liked: 1703 times
- Joined: Feb 04, 2013 2:07 pm
- Full Name: Dmitry Popov
- Location: Prague
- Contact:
Re: Performance of first synthetic full to tape?
Hello Paul,
To tell the truth synthesized full to tape should be create pretty fast. I suggest opening a support case and let the support team check the logs.
To tell the truth synthesized full to tape should be create pretty fast. I suggest opening a support case and let the support team check the logs.
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Ah, I hoped as much. I just realised when checking again that the job has actually run & eventually failed a few times with the following error:
Synthetic full
Have just logged as 01721125 but does this imply it can't load a tape, or perhaps it can't control the library? I did a library rescan which comes back ok...
Synthetic full
Code: Select all
Tape not exchanged. Failed to connect to Veeam Data Mover Service on host '<remote-tape-server>', port '6162' [<remote-tape-server>] Failed to open management RPC connection to proxy service. Port: '6162'. The remote procedure call failed RPC function call failed. Function name: [InvokerTestConnection]. Target machine: [<remote-IP>].
Failed to backup some full backup files, skipping the corresponding incremental backups
-
- Product Manager
- Posts: 14716
- Liked: 1703 times
- Joined: Feb 04, 2013 2:07 pm
- Full Name: Dmitry Popov
- Location: Prague
- Contact:
Re: Performance of first synthetic full to tape?
Thanks for sharing - lets await for the application log analysis as the behavior does not look like expected.
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Patch applied, seems nothing has changed for the job, running for 1hr so far, nothing written to tape & .vsb file is up to 2.5Gb so far.
I've uploaded latest logs to the call...
I've uploaded latest logs to the call...
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Looks like 2hrs is the fail point at which it gives up with the following:
[Have added the latest logs to the ticket]
Code: Select all
11/03/2016 12:51:23 :: Tape not exchanged.
Failed to connect to Veeam Data Mover Service on host '<tape/repos-server-name>', port '6162'
<tape/repos-server-name> Failed to open management RPC connection to proxy service. Port: '6162'.
The remote procedure call failed
RPC function call failed. Function name: [InvokerTestConnection]. Target machine: [<tape/repos-IP-Address>].
-
- VP, Product Management
- Posts: 7076
- Liked: 1510 times
- Joined: May 04, 2011 8:36 am
- Full Name: Andreas Neufert
- Location: Germany
- Contact:
Re: Performance of first synthetic full to tape?
Just a guess... Is your Tape Server on another server than B&R itself?
Do you use a special account to connect to this system. Maybe adding it to the User and roles will solve it.
As well take care that B&R can connect to the system without Windows Firewall stuff.
And another idea... check time on both systems. They need to be on exact GMT time (but can be different time zones).
Do you use a special account to connect to this system. Maybe adding it to the User and roles will solve it.
As well take care that B&R can connect to the system without Windows Firewall stuff.
And another idea... check time on both systems. They need to be on exact GMT time (but can be different time zones).
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Thanks for the useful suggestions. You're right in that it is another B&R server, however I actually created a much smaller tape job also directed to that server which has worked perfectly each time I ran it so that implies to me that all comms etc. is at it should be, there's just something "funky" with this particular job for some reason...
-
- VP, Product Management
- Posts: 7076
- Liked: 1510 times
- Joined: May 04, 2011 8:36 am
- Full Name: Andreas Neufert
- Location: Germany
- Contact:
Re: Performance of first synthetic full to tape?
If I remember correctly some RPC errors are fixed with v8 latest patch. Please implement if not done yet.
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Thanks also but at supports request I patched it last week to latest so most recent test was with the latest v8 build (will likely look to v9 about 2/3 weeks after its next patch is released)
-
- VP, Product Management
- Posts: 7076
- Liked: 1510 times
- Joined: May 04, 2011 8:36 am
- Full Name: Andreas Neufert
- Location: Germany
- Contact:
Re: Performance of first synthetic full to tape?
Antivirus without exclusions ? Veeam KB1999
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
(Replied to this a while ago, no idea why it isn't showing) No A/V on this server as it happens...
-
- Veteran
- Posts: 600
- Liked: 66 times
- Joined: Jun 13, 2013 10:08 am
- Full Name: Paul Kelly
- Contact:
Re: Performance of first synthetic full to tape?
Another support guy is now working on it and says that the log files for the synth full are showing trouble with communicating with the tape server, and yet even now I created another tape job (copying an endpoint copy to tape) which worked 100% fine on first go. :-/ Have a remote session scheduled for tomorrow morning now...
Who is online
Users browsing this forum: No registered users and 22 guests