Re: disk used percentage is not symmetric on datanodes (balancer)
On Mon, Mar 25, 2013 at 4:29 AM, Tapas Sarangi tapas.sara...@gmail.comwrote: Hi, Thanks for the explanation. Where can I find the java code for balancer that utilizes the threshold value and calculate it myself as you mentioned ? I think I understand your calculation, but would like to see the code. src/hdfs/org/apache/hadoop/hdfs/server/balancer/Balancer.java see BalancerDatanode If I set the threshold to 5 instead of 10, then the smaller nodes will have a maximum of 95% full where the larger nodes disk-usage will increase from 80% to 85%. Now my question to you and the experts is when I run the balancer, is the following command enough to set the threshold to a different value : hadoop balancer -threshold 5 yes Thanks to all for the suggestions... --- today i thought about my advice for you and i have understood that i wrong. for example we have 100 nodes where 80 with 12Tb and 20 with 72 Tb.all node have 10 Tb data. averege cluster dfs used 1000/2600*100=38.5 for 12Tb node dfs used it is 83.3 from capacity for 72Tb nodes its 13.9. node is balanced if averege cluster dfs used +threshold node dfs used averege cluster dfs used - threshold. data will move from 12Tb to 72 Tb and when 12Tb nodes will have 48.5 of capacity balancer will stop. In this time 72tb node have 36.1 % of capacity. the cluster will grow up,in ideal case when cluster dfs used capacity 90 % .72Tb nodes will about 80% of capacity and 12Tb have about 100 % capacity.After that you have about 288Tb freespace - On Sun, Mar 24, 2013 at 11:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For
Re: disk used percentage is not symmetric on datanodes (balancer)
Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVEif you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
-setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVEif you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVEif you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVEif you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog
Re: disk used percentage is not symmetric on datanodes (balancer)
you said that threshold=10.Run mannualy command : hadoop balancer threshold 9.5 ,then 9 and so with 0.5 step. On Sun, Mar 24, 2013 at 11:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVEif you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
Hi, Thanks for the idea, I will give this a try and report back. My worry is if we decommission a small node (one at a time), will it move the data to larger nodes or choke another smaller nodes ? In principle it should distribute the blocks, the point is it is not distributing the way we expect it to, so do you think this may cause further problems ? - On Mar 24, 2013, at 3:37 PM, Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb
Re: disk used percentage is not symmetric on datanodes (balancer)
On Mar 24, 2013, at 3:40 PM, Alexey Babutin zorlaxpokemon...@gmail.com wrote: you said that threshold=10.Run mannualy command : hadoop balancer threshold 9.5 ,then 9 and so with 0.5 step. We are not setting threshold anywhere in our configuration and thus considering the default which I believe is 10. Why do you suggest such steps need to be tested for balancer ? Please explain. I guess we had a discussion earlier on this thread and came to the conclusion that the threshold will not help in this situation. - On Sun, Mar 24, 2013 at 11:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned. -- http://balajin.net/blog http://flic.kr/balajijegan -- http://balajin.net/blog http://flic.kr/balajijegan
Re: disk used percentage is not symmetric on datanodes (balancer)
I think that it makes help,but start from 1 node.watch where data have moved On Mon, Mar 25, 2013 at 12:44 AM, Tapas Sarangi tapas.sara...@gmail.comwrote: Hi, Thanks for the idea, I will give this a try and report back. My worry is if we decommission a small node (one at a time), will it move the data to larger nodes or choke another smaller nodes ? In principle it should distribute the blocks, the point is it is not distributing the way we expect it to, so do you think this may cause further problems ? - On Mar 24, 2013, at 3:37 PM, Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For
Re: disk used percentage is not symmetric on datanodes (balancer)
It shouldn't cause further problems since most of your small nodes are already their capacity. You could set or increase the dfs reserved property on your smaller nodes to force the flow of blocks onto the larger nodes. On Mar 24, 2013 4:45 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Thanks for the idea, I will give this a try and report back. My worry is if we decommission a small node (one at a time), will it move the data to larger nodes or choke another smaller nodes ? In principle it should distribute the blocks, the point is it is not distributing the way we expect it to, so do you think this may cause further problems ? - On Mar 24, 2013, at 3:37 PM, Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster.
Re: disk used percentage is not symmetric on datanodes (balancer)
Thanks. Does this need a restart of hadoop in the nodes where this modification is made ? - On Mar 24, 2013, at 8:06 PM, Jamal B jm151...@gmail.com wrote: dfs.datanode.du.reserved You could tweak that param on the smaller nodes to force the flow of blocks to other nodes. A short term hack at best, but should help the situation a bit. On Mar 24, 2013 7:09 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: On Mar 24, 2013, at 4:34 PM, Jamal B jm151...@gmail.com wrote: It shouldn't cause further problems since most of your small nodes are already their capacity. You could set or increase the dfs reserved property on your smaller nodes to force the flow of blocks onto the larger nodes. Thanks. Can you please specify which are the dfs properties that we can set or modify to force the flow of blocks directed towards the larger nodes than the smaller nodes ? - On Mar 24, 2013 4:45 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Thanks for the idea, I will give this a try and report back. My worry is if we decommission a small node (one at a time), will it move the data to larger nodes or choke another smaller nodes ? In principle it should distribute the blocks, the point is it is not distributing the way we expect it to, so do you think this may cause further problems ? - On Mar 24, 2013, at 3:37 PM, Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way :
Re: disk used percentage is not symmetric on datanodes (balancer)
Yes On Mar 24, 2013 9:25 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks. Does this need a restart of hadoop in the nodes where this modification is made ? - On Mar 24, 2013, at 8:06 PM, Jamal B jm151...@gmail.com wrote: dfs.datanode.du.reserved You could tweak that param on the smaller nodes to force the flow of blocks to other nodes. A short term hack at best, but should help the situation a bit. On Mar 24, 2013 7:09 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: On Mar 24, 2013, at 4:34 PM, Jamal B jm151...@gmail.com wrote: It shouldn't cause further problems since most of your small nodes are already their capacity. You could set or increase the dfs reserved property on your smaller nodes to force the flow of blocks onto the larger nodes. Thanks. Can you please specify which are the dfs properties that we can set or modify to force the flow of blocks directed towards the larger nodes than the smaller nodes ? - On Mar 24, 2013 4:45 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Thanks for the idea, I will give this a try and report back. My worry is if we decommission a small node (one at a time), will it move the data to larger nodes or choke another smaller nodes ? In principle it should distribute the blocks, the point is it is not distributing the way we expect it to, so do you think this may cause further problems ? - On Mar 24, 2013, at 3:37 PM, Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.comwrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For
Re:Re: disk used percentage is not symmetric on datanodes (balancer)
if the balancer is not running ,or with a low bandwith and slow reaction, i think there may have a signatual unsymmetric between datanodes . At 2013-03-25 04:37:05,Jamal B jm151...@gmail.com wrote: Then I think the only way around this would be to decommission 1 at a time, the smaller nodes, and ensure that the blocks are moved to the larger nodes. And once complete, bring back in the smaller nodes, but maybe only after you tweak the rack topology to match your disk layout more than network layout to compensate for the unbalanced nodes. Just my 2 cents On Sun, Mar 24, 2013 at 4:31 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks. We have a 1-1 configuration of drives and folder in all the datanodes. -Tapas On Mar 24, 2013, at 3:29 PM, Jamal B jm151...@gmail.com wrote: On both types of nodes, what is your dfs.data.dir set to? Does it specify multiple folders on the same set's of drives or is it 1-1 between folder and drive? If it's set to multiple folders on the same drives, it is probably multiplying the amount of available capacity incorrectly in that it assumes a 1-1 relationship between folder and total capacity of the drive. On Sun, Mar 24, 2013 at 3:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, thanks for pointing, but I already know that it is completing the balancing when exiting otherwise it shouldn't exit. Your answer doesn't solve the problem I mentioned earlier in my message. 'hdfs' is stalling and hadoop is not writing unless space is cleared up from the cluster even though df shows the cluster has about 500 TB of free space. --- On Mar 24, 2013, at 1:54 PM, Balaji Narayanan (பாலாஜி நாராயணன்) bal...@balajin.net wrote: -setBalancerBandwidth bandwidth in bytes per second So the value is bytes per second. If it is running and exiting,it means it has completed the balancing. On 24 March 2013 11:32, Tapas Sarangi tapas.sara...@gmail.com wrote: Yes, we are running balancer, though a balancer process runs for almost a day or more before exiting and starting over. Current dfs.balance.bandwidthPerSec value is set to 2x10^9. I assume that's bytes so about 2 GigaByte/sec. Shouldn't that be reasonable ? If it is in Bits then we have a problem. What's the unit for dfs.balance.bandwidthPerSec ? - On Mar 24, 2013, at 1:23 PM, Balaji Narayanan (பாலாஜி நாராயணன்) li...@balajin.net wrote: Are you running balancer? If balancer is running and if it is slow, try increasing the balancer bandwidth On 24 March 2013 09:21, Tapas Sarangi tapas.sara...@gmail.com wrote: Thanks for the follow up. I don't know whether attachment will pass through this mailing list, but I am attaching a pdf that contains the usage of all live nodes. All nodes starting with letter g are the ones with smaller storage space where as nodes starting with letter s have larger storage space. As you will see, most of the gXX nodes are completely full whereas sXX nodes have a lot of unused space. Recently, we are facing crisis frequently as 'hdfs' goes into a mode where it is not able to write any further even though the total space available in the cluster is about 500 TB. We believe this has something to do with the way it is balancing the nodes, but don't understand the problem yet. May be the attached PDF will help some of you (experts) to see what is going wrong here... Thanks -- Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision,
Re: disk used percentage is not symmetric on datanodes (balancer)
2013/3/20 Tapas Sarangi tapas.sara...@gmail.com Thanks for your reply. Some follow up questions below : On Mar 20, 2013, at 5:35 AM, Алексей Бабутин zorlaxpokemon...@gmail.com wrote: dfs.balance.bandwidthPerSec in hdfs-site.xml.I think balancer cant help you,because it makes all the nodes equal.They can differ only on balancer threshold.Threshold =10 by default.It means,that nodes can differ up to 350Tb between each other in 3.5Pb cluster.If Threshold =1 up to 35Tb and so on. If we use multiple racks, let's assume we have 10 racks now and they are equally divided in size (350 TB each). With a default threshold of 10, any two nodes on a given rack will have a maximum difference of 35 TB, is this correct ? Also, does this mean the difference between any two racks will also go down to 35 TB ? Balancer know about topology,but when calculate balancing it operates only with nodes not with racks. You can see how it work in Balancer.java in BalancerDatanode about string 509. I was wrong about 350Tb,35Tb it calculates in such way : For example: cluster_capacity=3.5Pb cluster_dfsused=2Pb avgutil=cluster_dfsused/cluster_capacity*100=57.14% used cluster capacity Then we know avg node utilization (node_dfsused/node_capacity*100) .Balancer think that all good if avgutil +10node_utilizazation=avgutil-10. Ideal case that all node used avgutl of capacity.but for 12TB node its only 6.5Tb and for 72Tb its about 40Tb. Balancer cant help you. Show me http://namenode.rambler.ru:50070/dfsnodelist.jsp?whatNodes=LIVE if you can. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned.
Re: disk used percentage is not symmetric on datanodes (balancer)
2013/3/19 Tapas Sarangi tapas.sara...@gmail.com On Mar 19, 2013, at 5:00 AM, Алексей Бабутин zorlaxpokemon...@gmail.com wrote: node A=12TB node B=72TB How many A nodes and B from 200 do you have? We have more number of A nodes than B. The ratio of the number is about 80, 20. Note that not all the B nodes are 72TB, that's a max value. Similarly for A it is a min. value. If you have more B than A you can deactivate A,clear it and apply again. Apply what ? It may not be a choice for an active system and it may cripple us for days. I suppose that cluster about 3-5 Tb.Run balancer with threshold 0.2 or 0.1. You meant 3.5 PB, then you are about right. What this threshold does exactly ? We are not setting the threshold manually, but isn't hadoop's default 0.1 ? Different servers in one rack is bad idea.You should rebuild cluster with multiple racks. Why bad idea ? We are using hadoop as a file system not as a scheduler. How multiple racks are going to help in balancing the disk-usage across datanodes ? dfs.balance.bandwidthPerSec in hdfs-site.xml.I think balancer cant help you,because it makes all the nodes equal.They can differ only on balancer threshold.Threshold =10 by default.It means,that nodes can differ up to 350Tb between each other in 3.5Pb cluster.If Threshold =1 up to 35Tb and so on. In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. Why did you select hdfs?May be lustre,cephfs and other is better choise. -Tapas 2013/3/19 Tapas Sarangi tapas.sara...@gmail.com Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
Thanks for your reply. Some follow up questions below : On Mar 20, 2013, at 5:35 AM, Алексей Бабутин zorlaxpokemon...@gmail.com wrote: dfs.balance.bandwidthPerSec in hdfs-site.xml.I think balancer cant help you,because it makes all the nodes equal.They can differ only on balancer threshold.Threshold =10 by default.It means,that nodes can differ up to 350Tb between each other in 3.5Pb cluster.If Threshold =1 up to 35Tb and so on. If we use multiple racks, let's assume we have 10 racks now and they are equally divided in size (350 TB each). With a default threshold of 10, any two nodes on a given rack will have a maximum difference of 35 TB, is this correct ? Also, does this mean the difference between any two racks will also go down to 35 TB ? In ideal case with replication factor 2 ,with two nodes 12Tb and 72Tb you will be able to have only 12Tb replication data. Yes, this is true for exactly two nodes in the cluster with 12 TB and 72 TB, but not true for more than two nodes in the cluster. Best way,on my opinion,it is using multiple racks.Nodes in rack must be with identical capacity.Racks must be identical capacity. For example: rack1: 1 node with 72Tb rack2: 6 nodes with 12Tb rack3: 3 nodes with 24Tb It helps with balancing,because dublicated block must be another rack. The same question I asked earlier in this message, does multiple racks with default threshold for the balancer minimizes the difference between racks ? Why did you select hdfs?May be lustre,cephfs and other is better choise. It wasn't my decision, and I probably can't change it now. I am new to this cluster and trying to understand few issues. I will explore other options as you mentioned.
Re: disk used percentage is not symmetric on datanodes (balancer)
node A=12TB node B=72TB How many A nodes and B from 200 do you have? If you have more B than A you can deactivate A,clear it and apply again. I suppose that cluster about 3-5 Tb.Run balancer with threshold 0.2 or 0.1. Different servers in one rack is bad idea.You should rebuild cluster with multiple racks. 2013/3/19 Tapas Sarangi tapas.sara...@gmail.com Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
On Mar 19, 2013, at 5:00 AM, Алексей Бабутин zorlaxpokemon...@gmail.com wrote: node A=12TB node B=72TB How many A nodes and B from 200 do you have? We have more number of A nodes than B. The ratio of the number is about 80, 20. Note that not all the B nodes are 72TB, that's a max value. Similarly for A it is a min. value. If you have more B than A you can deactivate A,clear it and apply again. Apply what ? It may not be a choice for an active system and it may cripple us for days. I suppose that cluster about 3-5 Tb.Run balancer with threshold 0.2 or 0.1. You meant 3.5 PB, then you are about right. What this threshold does exactly ? We are not setting the threshold manually, but isn't hadoop's default 0.1 ? Different servers in one rack is bad idea.You should rebuild cluster with multiple racks. Why bad idea ? We are using hadoop as a file system not as a scheduler. How multiple racks are going to help in balancing the disk-usage across datanodes ? -Tapas 2013/3/19 Tapas Sarangi tapas.sara...@gmail.com Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
Any more follow ups ? Thanks -Tapas On Mar 19, 2013, at 9:55 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: On Mar 18, 2013, at 11:50 PM, Harsh J ha...@cloudera.com wrote: What do you mean that the balancer is always active? meaning, the same process is active for a long time. The process that starts may not be exiting at all. We have a cron job set to run it every 10 minutes, but that's not in effect because the process may never exit. It is to be used as a tool and it exits once it balances in a specific run (loops until it does, but always exits at end). The balancer does balance based on usage percentage so that is what you're probably looking for/missing. May be. How does the balancer look for the usage percentage ? -Tapas On Tue, Mar 19, 2013 at 6:56 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, On Mar 18, 2013, at 8:21 PM, 李洪忠 lhz...@hotmail.com wrote: Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. Like I mentioned earlier in my reply to Bertrand, we haven't considered rack awareness for the cluster, currently it is considered as just one rack. Can that be the problem ? I don't know… -Tapas 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux -- Harsh J
Re: disk used percentage is not symmetric on datanodes (balancer)
If your balancer does not exit, then it means its heavily working in iterations trying to balance your cluster. The default bandwidth allows only for limited transfer speed (10 Mbps) to not affect the cluster's RW performance while moving blocks between DNs for balancing, so the operation may be slow unless you raise the allowed bandwidth. On Wed, Mar 20, 2013 at 7:37 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Any more follow ups ? Thanks -Tapas On Mar 19, 2013, at 9:55 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: On Mar 18, 2013, at 11:50 PM, Harsh J ha...@cloudera.com wrote: What do you mean that the balancer is always active? meaning, the same process is active for a long time. The process that starts may not be exiting at all. We have a cron job set to run it every 10 minutes, but that's not in effect because the process may never exit. It is to be used as a tool and it exits once it balances in a specific run (loops until it does, but always exits at end). The balancer does balance based on usage percentage so that is what you're probably looking for/missing. May be. How does the balancer look for the usage percentage ? -Tapas On Tue, Mar 19, 2013 at 6:56 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, On Mar 18, 2013, at 8:21 PM, 李洪忠 lhz...@hotmail.com wrote: Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. Like I mentioned earlier in my reply to Bertrand, we haven't considered rack awareness for the cluster, currently it is considered as just one rack. Can that be the problem ? I don't know… -Tapas 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux -- Harsh J -- Harsh J
Re: disk used percentage is not symmetric on datanodes (balancer)
Thanks for the reply. How can I assign a new value to the transfer speed for the balancer ? Is this the parameter, dfs.balance.bandwidthPerSec ? Where should this go, in conf/hdfs-site.xml ? or conf/core-site.xml ? -Tapas On Mar 19, 2013, at 11:05 PM, Harsh J ha...@cloudera.com wrote: If your balancer does not exit, then it means its heavily working in iterations trying to balance your cluster. The default bandwidth allows only for limited transfer speed (10 Mbps) to not affect the cluster's RW performance while moving blocks between DNs for balancing, so the operation may be slow unless you raise the allowed bandwidth. On Wed, Mar 20, 2013 at 7:37 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Any more follow ups ? Thanks -Tapas On Mar 19, 2013, at 9:55 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: On Mar 18, 2013, at 11:50 PM, Harsh J ha...@cloudera.com wrote: What do you mean that the balancer is always active? meaning, the same process is active for a long time. The process that starts may not be exiting at all. We have a cron job set to run it every 10 minutes, but that's not in effect because the process may never exit. It is to be used as a tool and it exits once it balances in a specific run (loops until it does, but always exits at end). The balancer does balance based on usage percentage so that is what you're probably looking for/missing. May be. How does the balancer look for the usage percentage ? -Tapas On Tue, Mar 19, 2013 at 6:56 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, On Mar 18, 2013, at 8:21 PM, 李洪忠 lhz...@hotmail.com wrote: Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. Like I mentioned earlier in my reply to Bertrand, we haven't considered rack awareness for the cluster, currently it is considered as just one rack. Can that be the problem ? I don't know… -Tapas 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux -- Harsh J -- Harsh J
disk used percentage is not symmetric on datanodes (balancer)
Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas
Re: disk used percentage is not symmetric on datanodes (balancer)
And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.comwrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux
Re: disk used percentage is not symmetric on datanodes (balancer)
Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com mailto:tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com mailto:decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com mailto:tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux
Re: disk used percentage is not symmetric on datanodes (balancer)
On Mar 18, 2013, at 6:17 PM, Bertrand Dechoux decho...@gmail.com wrote: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. This else is probably what's happening. I just checked the logs. Its active almost all the time. What threshold is used? Don't know what's this. How can I find out ? About the small and big datanodes, how are they distributed with regards to racks? We haven't considered rack awareness for our cluster. It is currently considered as one rack. I am going through some docs to figure out how I can implement this after the upgrade. About files, how is used the replication factor(s) and block size(s)? This is 2. Surely trivial questions again. Not really :) Thanks -Tapas Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux
Re: disk used percentage is not symmetric on datanodes (balancer)
Hi, On Mar 18, 2013, at 8:21 PM, 李洪忠 lhz...@hotmail.com wrote: Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. Like I mentioned earlier in my reply to Bertrand, we haven't considered rack awareness for the cluster, currently it is considered as just one rack. Can that be the problem ? I don't know… -Tapas 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux
Re: disk used percentage is not symmetric on datanodes (balancer)
What do you mean that the balancer is always active? It is to be used as a tool and it exits once it balances in a specific run (loops until it does, but always exits at end). The balancer does balance based on usage percentage so that is what you're probably looking for/missing. On Tue, Mar 19, 2013 at 6:56 AM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, On Mar 18, 2013, at 8:21 PM, 李洪忠 lhz...@hotmail.com wrote: Maybe you need to modify the rackware script to make the rack balance, ie, all the racks are the same size, on rack by 6 small nodes, one rack by 1 large nodes. P.S. you need to reboot the cluster for rackware script modify. Like I mentioned earlier in my reply to Bertrand, we haven't considered rack awareness for the cluster, currently it is considered as just one rack. Can that be the problem ? I don't know… -Tapas 于 2013/3/19 7:17, Bertrand Dechoux 写道: And by active, it means that it does actually stops by itself? Else it might mean that the throttling/limit might be an issue with regard to the data volume or velocity. What threshold is used? About the small and big datanodes, how are they distributed with regards to racks? About files, how is used the replication factor(s) and block size(s)? Surely trivial questions again. Bertrand On Mon, Mar 18, 2013 at 10:46 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hi, Sorry about that, had it written, but thought it was obvious. Yes, balancer is active and running on the namenode. -Tapas On Mar 18, 2013, at 4:43 PM, Bertrand Dechoux decho...@gmail.com wrote: Hi, It is not explicitly said but did you use the balancer? http://hadoop.apache.org/docs/r1.0.4/commands_manual.html#balancer Regards Bertrand On Mon, Mar 18, 2013 at 10:01 PM, Tapas Sarangi tapas.sara...@gmail.com wrote: Hello, I am using one of the old legacy version (0.20) of hadoop for our cluster. We have scheduled for an upgrade to the newer version within a couple of months, but I would like to understand a couple of things before moving towards the upgrade plan. We have about 200 datanodes and some of them have larger storage than others. The storage for the datanodes varies between 12 TB to 72 TB. We found that the disk-used percentage is not symmetric through all the datanodes. For larger storage nodes the percentage of disk-space used is much lower than that of other nodes with smaller storage space. In larger storage nodes the percentage of used disk space varies, but on average about 30-50%. For the smaller storage nodes this number is as high as 99.9%. Is this expected ? If so, then we are not using a lot of the disk space effectively. Is this solved in a future release ? If no, I would like to know if there are any checks/debugs that one can do to find an improvement with the current version or upgrading hadoop should solve this problem. I am happy to provide additional information if needed. Thanks for any help. -Tapas -- Bertrand Dechoux -- Harsh J