10-14-2015 08:17 AM
Hello
I have a question regarding deduplication and AIR : when a replication is interrupted for any reason (link failure, server reboot, ....) are the block already sent to the other site totaly lost ? With other words, when the replication is restarted (from 0 of course) , is it really not optimized (supposing it is the first time it runs) or the first part of the replication can take advantage of already sent blocks ?
I imagine the blocks received by the remote MSDP are stored then could be potentially be used ... but maybe they are flagged "to be deleted" when the replciation process is broken and then can not be used ...
If anybody has knowlegde or ideas on this point ....
Best regards
Solved! Go to Solution.
10-19-2015 04:05 AM
Hello,
refer to GarbageCheckRemainDCCount parameter in Dedup Guide:
The number of containers from failed jobs not to check for garbage. A
failed backup or replication job still produces data containers. Because
failed jobs are retried, retaining those containers means NetBackup
does not have to send the fingerprint information again. As a result,
retried jobs consume less time and fewer system resources than when
first run.
Rgds
10-16-2015 05:00 PM
*bump* (...because I like this question... does anyone know?)
10-17-2015 02:23 AM
Only way we have found is to run another backup that has replication too, this seems to restart the broken replication.
Please vote for the replication retry idea here:
10-19-2015 04:05 AM
Hello,
refer to GarbageCheckRemainDCCount parameter in Dedup Guide:
The number of containers from failed jobs not to check for garbage. A
failed backup or replication job still produces data containers. Because
failed jobs are retried, retaining those containers means NetBackup
does not have to send the fingerprint information again. As a result,
retried jobs consume less time and fewer system resources than when
first run.
Rgds
10-24-2015 06:29 AM
Thanks a lot for this information !
Does this mean that failed/retried jobs are keeping some DCs, but cancelled ones do not ?
10-24-2015 06:40 AM
A cancelled job is a failed job. I would expect the same DC / garbage rules to apply to any failed job no matter what the status is. I would expect the garbage DCs to only exist (be retained) for a specific length of time.
What is not clear to me is...
1) Are all garbage DCs deleted when garbage DC expiry runs no matter how old they are?
2) Or, are garbage DCs aged, and when garbage DC runs is it only those garbage DCs that are over a certain age which are deleted?