Hi Martin,
Any update with the .snar files?
Can you let me know how to patch foreman-maintain with option --no-check-device
Thanks
Farista
Hi Martin,
Any update with the .snar files?
Can you let me know how to patch foreman-maintain with option --no-check-device
Thanks
Farista
Hi Farista,
I analyzed the .snar
files you provided. From those it seems that all the mongo content changes prior the first backup in the day. If you look at the data attached you can see what is packed in the .snar
file. What is interesting is content of the dirents
entries. It is basically list of filenames in the directory prefixed with either ‘Y’ if it changed since last backup or ‘N’ if not. In the data you’ve sent only second backup on Wednesday shows that “almost nothing” changed since last backup. You can check stats on the files in the directory to confirm (all backuped files should be modified or changed since previous backup.
From these findings I assume foreman-maintain and also tar works correctly. Just the mongo with WiredTiger engine might not be ideal target for incremental backups. BTW what is the difference in size of the mongo_data.tar.gz in the wednesdays backups? According to your .snars all the content in the journal/
was backed up in both runs. This directory takes about 95% of the mongo size (unpacked) on my setup.
Do you see the same behavior in other tarballs in the backup or just in mongo?
I also can confirm your device ids keep preserved between backups so it makes no sense to add the --no-check-device
.
Regards,
Martinsnars_in_txt.tar (30 KB)
Hi Martin,
no-check-device should have no impact in taking incremental backup, since the ids are preserved.
Anyways we can give it a try this option as well.
Can you please make a command to patch foreman-maintain and add --no-check-device ?
Thanks
Farista Singh