Imported backups and catalog replication to EM

[StanO]

I searched the boards a bit before attempting to populate the Enterprise Manager catalog from imported backups that previously failed to replicate their catalogs to the EM server (I allowed the disk to fill - shame on me). I didn't find a definitive answer on if this would work as expected so I'm following up here in hopes of helping others that find themselves in the same predicament.

Good news - after importing a backup (and importing the guest file catalog), having the catalog replicated to the Enterprise Manager is as easy as manually starting a collection from the VBREM management console (waiting for a sheduled or triggered collection appears to also do the job). Kudos to Veeam for having this foresight.

I do have a question though about fully recovering from this situation (or recovering as much as possible). I am unable to recover the catalogs for some of the backups as my retention period on tape has been exceeded for some of the daily backups. There exists in the catalog on the EM server some folders for the days where the disk was full that only have partial contents (small files and folders made it to disk, but the guestindexdata.zip file didn't). Searches that include dates of these partial catalogs throw some errors. Is it acceptable to delete the folders related to these days in the catalog on the EM (in both the Machines and Sessions folders) to clear this condition? Put another way, has anybody selectively deleted folders from the catalog on the EM and have evidence that it does not sacrifice the integrity of the catalog as a whole?

Also, while I accept all the fault for allowing the disk to fill, I'm not too proud to make a feature request of having the EM send notifications about a disk full condition.

Lastly, it appears that using NTFS compression has become accepted standard practice since the last time I visted the subject, so barring any warnings or horror stories in response to this post I'm going to move forward with enabling it.

Thanks,

Stan

[Vitaliy S.]

Hello Stan,

I didn't find a definitive answer on if this would work as expected so I'm following up here in hopes of helping others that find themselves in the same predicament. Good news - after importing a backup (and importing the guest file catalog), having the catalog replicated to the Enterprise Manager is as easy as manually starting a collection from the VBREM management console (waiting for a sheduled or triggered collection appears to also do the job). Kudos to Veeam for having this foresight.

Yes, catalog data is contained inside the backup file itself, so you also import this data when importing backup. In other words, you will not lose your guest file indexes if you lose VBRCatalog.

Put another way, has anybody selectively deleted folders from the catalog on the EM and have evidence that it does not sacrifice the integrity of the catalog as a whole?

Yes, it is safe to delete indexes you no longer need, see this topic for more info: Guest File System Indexing

Also, while I accept all the fault for allowing the disk to fill, I'm not too proud to make a feature request of having the EM send notifications about a disk full condition.

Thanks for the feature request, however can you please tell me where EM is located at? Is it a VM? Do you have any monitoring solution (for ex. Veeam ONE), that can track, monitor and alert on free space of the Guest OS disks?

Lastly, it appears that using NTFS compression has become accepted standard practice since the last time I visted the subject, so barring any warnings or horror stories in response to this post I'm going to move forward with enabling it.

Yes, enabling NTFS compression is a common practice among our existing customers.

Thank you!

[StanO]

Vitality:

Thank you for the prompt reply and comprehensive addressing of all my questions.

Our VBREM server is a VM, however we do not currently have a solution that monitors disk space usage across our virtual infrastructure. I'll consider it for next year's budget. It would be nice, however if VBR could use the notification components already inside the product to give indication when an internal process such as catalog replication fails.

Thanks,

Stan

[Vitaliy S.]

If there is no budget now, you can use the free edition of Veeam ONE that will address this issue for you as well.

[StanO]

Vitaliy:

Thanks for the pointer, I'll give it a look.

Apologies for my misspelling (or just outright wrongness) of your name.

Stan

[StanO]

A follow up on this.

Unfortunately my adventure with recovering the catalogs continues. I discovered that although importing a backup populates the catalog data on the EM server, the catalog data stays on the EM only as long as the imported backups remain imported. The next catalog sync after the imported jobs are removed from the backups causes their corresponding catalogs to be deleted from the EM. I'm assuming it is a catalog sync operation and not the removing of the imported backups as I didn't notice until I thought I was almost done with my catalog recoveries that this was happening.

It appears that changing the line in the refs.txt file in the machine catalog folder from

backup.0.IsBackupImported=True
to
backup.0.IsBackupImported=False

will allow the individual machine catalog data for a session to remain on the EM through a catalog sync operation and searches appear successful. There is some additional data in the refs.txt file for jobs which have their catalog data on the EM through normal measures, but this data looks largly inconsequential for the purpose of having a searchable catalog that will at least tell you if a file by a specific name existed on a server at a specific point in time. This data normally looks like this:
backupSession.0.BackupServer=VEEAMBACK-01
backupSession.0.JobName=File Servers 01
backupSession.0.SessionDateUtc=02/08/2013 01:00:19.22
backupSession.0.OibOrigin=New

Editing the refs.txt file in the catalog is likely unsupported, but it does appear to do the job. If I encounter problems in the next 103 weeks (when the modified catalog entries will expire) I'll follow up here.

Stan

[foggy]

Stan, you're right in that manual editing of this file is neither recommended nor supported. There is a chance that indexes of the removed backup will not be deleted according to retention at all (if there is no corresponding backup session exist in Veeam B&R).

[StanO]

Taking a step back, I've conceived of another way I can make these catalogs available on a semi-permanent basis. I am able to restore from the backups of our Veeam Backup Server (which is a physical machine) the copies of the catalogs as they existed on the backup server during the week that the Veeam backup sets were on disk.

Is there any supported way of getting these recovered catalogs to my semi-permanent catalog repository on the EM?

I've been successful in just copying the recovered catalogs into the VBRCatalog\Index\Machines folder on the EM for the purpose of searching them, so I know I can set them aside in their own directory to use if the need arises to search the specific period of time when the catalog replication failed. I would like to know if there is a way to have them added and persist in the EM Catalog as they normally would have been so two years from now they will be purged via the catalog retention policy.

Thanks,

Stan

[foggy]

Stan, I won't say it is supported either as it requires some manual intervention in normal operation of the product components. However, you could copy the recovered catalog data on EM and this should allow you to search in corresponding backups. Make sure you copy both Machines and Sessions folder entries for the required time period. Again, I cannot guarantee you that this tweak gives the same result as proper catalog replication in case where everything worked fine (no replication failure), but at least you can do this safely and see whether this helps.