Custom Query (684 matches)

Until now, we used the NFS remote resource as a backup device, which was mounted on the target servers only for the time of the backup task. At the same time, we used the '-N' option with the 'mondoarchive' command to automatically exclude all mounted remote file systems, including the NFS resource with backups. This worked fine.

But recently, to reduce backup window time, we migrated to use Dell EMC DD BoostFS (boostfs) in conjunction with Dell EMC DD9300HA storage. And after that, the above option '-N' certainly stopped working, since the remote BoostFS resource with backups is not mounted via 'fstab' and looks like this:

=== $ df -hT Filesystem Type Size Used Avail Use% Mounted on ... boostfs fuse.boostfs 202T 82T 120T 41% /mnt/nfs_backups ===

The BoostFS remote resource is mounted at the time of backup task with a simple command:

# /opt/emc/boostfs/bin/boostfs mount /mnt/nfs_backups

Therefore, I really ask the developers to add support for the Dell EMC DD BoostFS (boostfs) remote file system for the above '-N' option. If required, I can send (privately) detailed 'mondoarchive' runtime log-files.

My working versions: -- Mondo Archive v3.3.0-r3697M (RHEL/CentOS-6) -- Mondo Archive v3.3.0-r3699 (RHEL/CentOS-7)

afio is missing from Debian. We need to make the dep based on star then

I found a bug in src/lib/mr_str.c. The following line:

q = outstr + strlen(outstr) -1;

will cause a SIGABRT on latest rhel 7. See parts of the log:

*** Error in `/usr/sbin/mondoarchive': corrupted size vs. prev_size: 0x000000000065b5d0 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x7f7c4)[0x7ffff765b7c4]
/lib64/libc.so.6(+0x82fd4)[0x7ffff765efd4]
/lib64/libc.so.6(+0x840c2)[0x7ffff76600c2]
/lib64/libc.so.6(realloc+0x1d2)[0x7ffff7662162]
/lib64/libc.so.6(getdelim+0x10b)[0x7ffff764b91b]
/usr/sbin/mondoarchive[0x4380bf]
/usr/sbin/mondoarchive[0x41e765]
/usr/sbin/mondoarchive[0x4291c8]
/usr/sbin/mondoarchive[0x402e90]
/lib64/libc.so.6(__libc_start_main+0xf5)[0x7ffff75fe505]
/usr/sbin/mondoarchive[0x4039b3]
==
....

Running it through valgrind, displays:

I think you have a EFI/UEFI partition.
==2587== Invalid read of size 1
==2587==    at 0x437DAA: mr_strip_spaces (mr_str.c:185)
==2587==    by 0x41E6FC: run_program_and_log_output (libmondo-fork.c:317)
==2587==    by 0x42B079: some_basic_system_sanity_checks (libmondo-tools.c:1016)
==2587==    by 0x42B4A3: pre_param_configuration (libmondo-tools.c:684)
==2587==    by 0x402E46: main (mondoarchive.c:328)
==2587==  Address 0x5ed05bf is 1 bytes before a block of size 1 alloc'd
==2587==    at 0x4C29EA3: malloc (vg_replace_malloc.c:309)
==2587==    by 0x537FF2F: __vasprintf_chk (vasprintf_chk.c:80)
==2587==    by 0x437FFA: UnknownInlinedFun (stdio2.h:210)
==2587==    by 0x437FFA: mr_asprintf_int (mr_mem.c:62)
==2587==    by 0x437D98: mr_strip_spaces (mr_str.c:181)
==2587==    by 0x41E6FC: run_program_and_log_output (libmondo-fork.c:317)
==2587==    by 0x42B079: some_basic_system_sanity_checks (libmondo-tools.c:1016)
==2587==    by 0x42B4A3: pre_param_configuration (libmondo-tools.c:684)
==2587==    by 0x402E46: main (mondoarchive.c:328)
==2587==

The following change fixes it:

q = outstr + strlen(outstr);

After that archive creation successfully completes.