Hi,
We noticed some really strange behaviour with cbrestorewrapper when trying to restore a bucket with around 40 Million Documents on an empty bucket. Launching it with -x rehash=1 -b bucketname results in a short period of heavy read utilisation ( backup files are read ) and then - nothing really happens.
Documents are pouring in at a pace of about 1 document in a minute or so. IO and CPU utilization are near zero. RAM uses a quarter of server capacity. (I try to import the data on a single node. )
At the same time we have a small bucket with around 70k docs which got restored nearly instantly.
Anyone knows what is going on here and how i can fix this? (CB 4.5)
thank you very much for your help - i really appreciate your hard work.
Regards,
Leon
Hi Leon
Could you please retry with the ‘–verbose’ flag and post the output?
Hi Will,
I could tackle the issue… i created the target bucket programatically before launching the import ( shellscript ).
Even though the creation process finished successful the bucket didn’t work correctly yet. if i launch import immidiatly after creation the whole thing is extremly slow - if i wait 30 seconds everything works like expected.
I’m not sure why this happened but it should be easy to reproduce.
Best Regards and Thank You,
Leon
The cbtransfer suite of tools (which includes cbbackup, cbrestore, cb*wrapper and others) is not tolerant to changes to the cluster map and bucket states which will occur shortly after bucket creation. In that respect this is expected behaviour (ie. that it doesn’t work).
If you need that kind of stability then I would recommend using the enterprise backup/restore tool which is designed with changes to the cluster map / bucket states in mind.
Thank you very much! Good to know - is this somewhere documented? Otherwise it would be great to add it to cbrestore docs. Like the thing with rehash.