Anytime you copy sstables around (sstableloader is excluded here), make sure 
you don't copy sstables from 2 nodes to the same directories without checking 
that the names don't collide or you'll lose one of the sstables


-- 
Jeff Jirsa


> On Aug 2, 2017, at 10:38 AM, Harika Vangapelli -T (hvangape - AKRAYA INC at 
> Cisco) <hvang...@cisco.com> wrote:
> 
> Jeff,  what is the meaning of this line you mentioned in below email..
> ‘making sure you dont overwrite any sstables with the same name’
>  
> <image001.png>
>  
> Harika Vangapelli
> Engineer - IT
> hvang...@cisco.com
> Tel:
> Cisco Systems, Inc.
> 
> 
> 
> United States
> cisco.com
> <image002.gif>Think before you print.
> This email may contain confidential and privileged material for the sole use 
> of the intended recipient. Any review, use, distribution or disclosure by 
> others is strictly prohibited. If you are not the intended recipient (or 
> authorized to receive for the recipient), please contact the sender by reply 
> email and delete all copies of this message.
> Please click here for Company Registration Information.
>  
> From: Jeff Jirsa [mailto:jji...@gmail.com] 
> Sent: Tuesday, August 1, 2017 5:13 PM
> To: cassandra <user@cassandra.apache.org>
> Subject: Re: Cassandra Data migration from 2.2.3 to 3.7
>  
> Roughly that approach should work, modulo copying data in/out to the right 
> paths, and making sure you dont overwrite any sstables with the same name.
> 
> (You dont need to refresh after the upgradesstables unless you've got 2 3.7 
> clusters, and you may need to refresh before the upgradesstables if the 3.7 
> cluster in step 3 is already running)
>  
>  
> On Tue, Aug 1, 2017 at 3:44 PM, Harika Vangapelli -T (hvangape - AKRAYA INC 
> at Cisco) <hvang...@cisco.com> wrote:
> Jeff, I tried the below steps for just 3 rows of data, It looks to be 
> working. But Confirm me this is right approach.
> 1.       Taking snapshot (using nodetool snapshot) in the existing cluster 
> (2.2 cluster)
> 
> 2.       Bring the snapshot to new cluster and copy in keyspce/table 
> (2.2->3.7)
> 
> 3.       Nodetool upgrade sstbales on new cluster (3.7)
> 
> 4.       Use noedtool refresh -- <keyspace> <table> (3.7)
> 
> Does this approach work?
>  
> <image001.png>
>  
> Harika Vangapelli
> Engineer - IT
> hvang...@cisco.com
> Tel:
> Cisco Systems, Inc.
> 
> 
> 
> United States
> cisco.com
>  
> <image002.gif>Think before you print.
> This email may contain confidential and privileged material for the sole use 
> of the intended recipient. Any review, use, distribution or disclosure by 
> others is strictly prohibited. If you are not the intended recipient (or 
> authorized to receive for the recipient), please contact the sender by reply 
> email and delete all copies of this message.
> Please click here for Company Registration Information.
>  
> From: Jeff Jirsa [mailto:jji...@gmail.com] 
> Sent: Tuesday, August 1, 2017 3:36 PM
> 
> To: user@cassandra.apache.org
> Subject: Re: Cassandra Data migration from 2.2.3 to 3.7
>  
> You can't stream across versions, so you either need to use the native 
> protocol (do writes with the normal INSERT operations) or you have to get to 
> the right sstable version
>  
> That means one of the three of these: 
>  
> - export to CSV (copy to/copy from, or Brian Hess' Cassandra-loader on github)
> - export to CSV and use CQLSSTableWriter to generate sstables you can bulk 
> load
> - create a minimal cluster, load the sstables in, and run upgradesstables to 
> rev the version and then stop that cluster and stream them into the new 
> cluster
> 
> 
> -- 
> Jeff Jirsa
>  
> 
> On Aug 1, 2017, at 3:30 PM, Harika Vangapelli -T (hvangape - AKRAYA INC at 
> Cisco) <hvang...@cisco.com> wrote:
> 
> Jeff, Thanks for initial pointer. I want to add some more information to it.
>  
> This is to migrate from one cluster (2.2 version) to 3.7 . The client who is 
> in 2.2 cluster they don’t need down time.
>  
> So How the data is migrated to new cluster without down time with different 
> versions?
>  
> Thanks,
> Harika
>  
> <image001.png>
>  
> Harika Vangapelli
> Engineer - IT
> hvang...@cisco.com
> Tel:
> Cisco Systems, Inc.
> 
> 
> 
> United States
> cisco.com
>  
> <image002.gif>Think before you print.
> This email may contain confidential and privileged material for the sole use 
> of the intended recipient. Any review, use, distribution or disclosure by 
> others is strictly prohibited. If you are not the intended recipient (or 
> authorized to receive for the recipient), please contact the sender by reply 
> email and delete all copies of this message.
> Please click here for Company Registration Information.
>  
> From: Jeff Jirsa [mailto:jji...@gmail.com] 
> Sent: Tuesday, August 1, 2017 2:48 PM
> To: user@cassandra.apache.org
> Subject: Re: Cassandra Data migration from 2.2.3 to 3.7
>  
> If you upgrade in place:
>  
> Stop each node one at a time
> Swap binaries
> Start each node
> Run upgradesstables when all nodes are bounced
>  
> If you want to take data from 2.2 and put it into a 3.x cluster, you'll need 
> to run upgradesstables on it offline somehow (like clone that 2.2 
> cluster/schema and run it offline with 3.x upgradesstables), then 
> sstableloader it into place
>  
> Note that 3.7 is probably not the version you want - 3.11 is probably much 
> more stable and solves some serious bugs (or 3.0.14 if you don't need 
> features from 3.7 ). The one exception to this would be if you're using a 3.7 
> lts branch (like instaclustr had at one point - and may still have). 
>  
> 
> -- 
> Jeff Jirsa
>  
> 
> On Aug 1, 2017, at 2:24 PM, Harika Vangapelli -T (hvangape - AKRAYA INC at 
> Cisco) <hvang...@cisco.com> wrote:
> 
> Please help me understand data migration path from 2.2.3 to 3.7.
>  
> 1.       Tried to take snapshot and load the snapshot using sstableloader; 
> but it is throwing Streaming Exception.
> 
> Please advise.
>  
> Thanks,
> Harika
>  
>  
> <image001.png>
>  
> Harika Vangapelli
> Engineer - IT
> hvang...@cisco.com
> Tel:
> Cisco Systems, Inc.
> 
> 
> 
> United States
> cisco.com
>  
> <image002.gif>Think before you print.
> This email may contain confidential and privileged material for the sole use 
> of the intended recipient. Any review, use, distribution or disclosure by 
> others is strictly prohibited. If you are not the intended recipient (or 
> authorized to receive for the recipient), please contact the sender by reply 
> email and delete all copies of this message.
> Please click here for Company Registration Information.
>  
>  

Reply via email to