cancel
Showing results for 
Search instead for 
Did you mean: 

ALL_LOCAL_DRIVES generating 71 for one stream.

PaulStevens
Level 4
Partner

Master server is appliance 5240 on 3.2/8.2 and is acting as media server on this policy.  Client is also 8.2

nyxalbnbu001Linux(3.10.0-957.62.1.el7.x86_64)UNIXMaster Server8.2ConnectedOff
nyxsvlalb444pLinux(3.10.0-1160.31.1.el7.x86_64)UNIXClient8.2ConnectedOff

 

Policy has Backup Selection   ALL_LOCAL_DRIVES  as the only selection.

Backing up to MSDP on the master

Full and Differential jobs are generating a 71 for a particular directory.    /run/user/13276/gvfs

In the detailed status I see 

Jul 15, 2021 8:58:49 AM - Error bpbrm (pid=211980) socket close failed, Bad file descriptor (9)
5 REPLIES 5

X2
Moderator
Moderator
   VIP   

@PaulStevens status 71 - none of the files in the file list exist [Status code reference]

And /run/user/13726/gvfs indicates a user-space mounted device on the server (gnome virtual file system). Is the file/directory still available on the file system? What do you get when do ls -l /run/user/13726/gvfs ? Secondly, do you need to backup that location?

 

Hamza_H
Moderator
Moderator
   VIP   

We had kind of the same problem and as mentioned in the reply above "/run/user/13726/gvfs indicates a user-space mounted device on the server (gnome virtual file system)" and I don't think you can backup this folders.. therefore , you must exclude them if you don't want to get these errors.

 

Link the the post : https://vox.veritas.com/t5/NetBackup/Backup-of-a-Redhat-server-is-partial-ERR-Cannot-lstat-gvfs-Errn...

I forgot to put the solution then, I have just added it now/

Hi @PaulStevens 

You will want to add an exclude for that "mount" but as it changes you will need a wildcard exclusion. 

When you use ALL_LOCAL_DRVES, NetBackup uses the bpmount utility to determine all the local file systems. There is nothing under the /run that is really needed to be backed up as it is all transient data - the directory contains system information data describing the system since it was booted. 

To exclude you will need to put something like "/run/user/*/gvfs" in the exclude_list file on the client as the number will change (depending on who is logged in), it is a specific mount point (one per user logged in) and is unreadable by ordinary means (lookup gvfsd-fuse if you want more details). 

Cheers
David

I do not have access on that server to create the exclude list.  I did however, see in a tech note that the problem occurs with ALL_LOCAL_DRIVES combined with Allow Multiple Data Streams.  I turned off Allow Multiple Data Streams and I got a successful backup.  Since the systems in this policy are small, I did not see a noticeable increase in the backup time.

Hi @PaulStevens 

You should be able to use the bpgetconfig and bpsetconfig commands to retrieve and set an exclude list on Linux/Unix hosts without login access to the client. First I would recommend getting what is currently there (which may well be nothing - but just in case). Then edit the file and push it back.

# bpgetconfig -e /tmp/exclude.tmp <client>
# vi /tmp/exclude.tmp
# bpsetconfig -h <client> -e /tmp/exclude.tmp

If you want to restrict the exclude_list to a particular policy and/or schedule add the "-c <policy>" and "-s <schedule>"  options to the bpsetconfig command - refer to the commands reference guide for more details.

David